@CoinJunkie said:
The conclusion of the article is not that "sniping doesn't work":
In one auction of each pair we submitted an early bid, and in the other we submitted the same bid exactly ten seconds before its close. Our results, from a set of 70 such pairs, indicate no benefit to sniping: we found evidence of 2.54% lower prices for the sniped auctions, but we did not find this benefit to be statistically significant.
In fact, sniping results in slightly lower prices, which is a benefit in spite of what the authors state above. From my perspective, the only downside to sniping is that in a multiway (eBay style) auction, if you are the last person to enter a bid, you can potentially not win the auction even though your bid was the highest. That would be due to your bid not being at least a full increment higher than the current price at the time it was entered.
This is a completely ignorant statement. When the difference is not statistically significant, it means that there is effectively no difference as it is within the margin of error of the measurment.
Use common sense! There is no possible way that sniping can cost you money, but it can save you money. Them's the facts. I don't care about "statistical significance". As I stated, the downside of sniping is that you will very occasionally fail to secure a lot when you in fact submitted the highest bid, so that's the only risk in it.
That's not true. If I submit a nuclear snipe bid to try to win it - which people do - then you can end up with a higher bid than you would normally submit. You actually can't afford to be conservative, because you'll have no chance to counter.
If I submit the same bid as I normally would, no matter what, you are correct. But then I'm less likely to win because the first one to the winning number wins.
Of course, it's not my study, so I don't feel too much passion to defend the conclusion. There are other studies that show slightly different outcomes. None of them seems to show a significant price difference or a significant difference in the likelihood of winning with the sole exception [of studies I found] of a study that showed that if you found unpopular auctions with few bidders you tended to do better. Although, in that case, is the success due to sniping or simple lack of competition?
Nonetheless, I find it interesting. You may not. So be it. But please argue with data not conjecture. Thank you.
People also submit nuclear bids early in auctions. I've seen it happen many times. Just as I've seen many snipe bids of the non-nuclear variety. If you want to turn this into a debate on whether snipe bidders are likely to bid more aggressively than non-snipers, I have no interest. From the paper you cited:
In one auction of each pair we submitted an early bid, and in the other we submitted the same bid exactly ten seconds before its close.
So I believe it was reasonable to assume that that was the scenario under discussion.
And for the record, I am in no way "anti-science".
Your words: ** I don't care about "statistical significance". **
Even if they submit the identical bid early and late, since they don't always win at their maximum bid, it is still possible that the snipe bids ended at a higher number. That, of course, is NOT what the data says. The data, however, also indicates that it didn't end at a lower number.
OK, let me rephrase what I assumed was obvious: I don't care about statistical significance in the context of this argument. I'm not going to rehash everything I've already written. But to summarize: Sniping (vs. making the same bid earlier) can never cost you money, but it can save you money. No statistics are required to prove that argument.
Again, that is only true relative to your max bid. You can't ever pay more than your max bid, sure. But you could make the same case about bidding early. Bidding first can never cost you money, but it could save you money.
>
It is a completely different situation if you are talking about the price paid rather than the max bid.
If bidding $100 on day one leads to an average purchase price of $80 and sniping at $100 leads to an average purchase price of $90 - assuming the type of bid is the ONLY difference - then sniping costs you $10.
Your point is only correct in the limited and moot context of the max bid. No one EVER pays more than their max bid, no matter when placed.
You are assuming that everyone's bid, at any time, is not influenced by what's already happened. You are assuming that everyone is always going to have the same max bid. That may not be true, including the sniper. If you calculate your snipe bid based on the price on day 6 rather than the opening bid on day 1, it may end up higher. There's a whole psychology to this.
There are so many logical fallacies and misdirections in this post, I'm not sure where to begin.
For the purposes of this argument, we are talking about the methodology in the cited study. Namely, the same pre-determined maximum bid is either placed early or at the last second. There is no psychology or emotion involved there. Agree or disagree?
If you want to stick to the methodology in this study, fine. Although your statement seemed more general. But your point is still moot. As I pointed out. If you bid $100 on day one and your win price is $80 and you bid THE SAME $100 on the final day and your win price is $90. And the only difference is the timing of the bid, then bidding late cost you $10. It IS POSSIBLE, contrary to your assertion, for snipe bidding to cost you money if you are looking at price paid because...
You can NOT ignore the psychology of the other bidders even if there is no psychology to your identical $100 bids. As you mention here...
Now consider the psychology of the other bidders in the auction. If you know anything about poker and game theory, then you know that the less your opponent knows about your intentions, the more advantageous for you. By bidding at the last possible moment, you deprive the opponent(s) of the time to react to your bid. By bidding at the last second, you also eliminate the possibility of shills probing your early bid with incremental raises.
Your "opponent" does not know what your max bid is if you bid on day one. Yes, you eliminate the possibility of shills probing your early bid, assuming that's even common. But you might also consider that you change the whole nature of the auction by eliminating weaker hands early. Overall activity is, according to another study, a factor in where snipers tend to snipe.
You also need to consider the increments. Let's say your "opponent" will bid no higher than $80. If you bid early and get to $80 first, he does not place another bid. On the other hand, if he places his $80 bid first, your $100 snipe forces you to pay $82 or $85 whatever the bid increment is. That is a very simple way that the snipe would cost you MORE even with identical behavior from all bidders.
None of this is an argument against the findings of the study, which are what they are.
Why don't you just admit that you either misspoke or were incorrect? First, you suggested that the statistically insignificant difference was somehow significant. When that didn't work, you suggested that there is no way you ever pay more with a snipe than an early bid, which I've also demonstrated is simply untrue.
It's okay, we still love you even if you're wrong.
I did not misspeak. You keep coming up with off the wall scenarios to try and save face. For example, if I bid my max early, I do not force out weaker hands unless someone else has already entered a bid higher than what they (the weaker hands) wanted to bid. In which case, the weaker hands were never going to be a factor in the auction anyway.
The only point of yours I agree with is that if you bid the same exact amount as someone else, it is advantageous to have done so first. That is probably why experienced snipers tend to add a small random amount to their max bid rather than use a round number.
None of that is off the wall.
Why do i have to save face? It's not even my study you are arguing with? And you just admitted that you could end up paying more with a snipe due to the increments. So, if nothing else you are wrong on that. And, frankly, the statistics speak for themselves.
But, that's okay, I don't ever expect you to admit that you were wrong...especially if it simultaneously means I was right.
The forum may judge. I am done.
You composed the title ("bid sniping does not work"). Yet the data disclosed in the study showed that snipers saved money on average. Then the authors conclude that the savings weren't statistically significant. OK, so don't publish the damn article until you've run enough trials to achieve statistical significance! By the same token, they certainly didn't prove that sniping costs the bidder money. At worst, it's inconclusive.
I did not admit that you'll end up paying more with a snipe. You'll end up winning slightly less often, but it is impossible to construct a scenario where given the same max bids placed by the other bidders you end up paying more with a snipe than had you bid the same amount first. I dare you to try.
NO, the conclusion of the study was quite the opposite: there were NO STATISTICALLY SIGNIFICANT SAVINGS.
As for delaying publication until you have a statistically significant difference ASSUMES that they would eventually find one. Even if they run 1000 trials, there will still be a statistical test of significance. While the confidence interval will be more narrow, it is never zero. And, if there is no statistically significant difference, there is no statistically significant difference.
He's correct here, @CoinJunkie. In statistics, the lack of a statistically significant result is, itself, significant. What it's saying is that although the numbers are different (sniping saves you money) given the parameters of the trial, the difference is within the error band, and the result could have just as easily said sniping costs you money. This is the same reason that medical trials may say a drug is no better than a placebo even when more patients who received the drug get better. If not enough more get better, the results may be due to randomness (the patients who got the drug were slightly less sick than those who got the placebo) rather than because the drug actually worked.
I agree, in the abstract. However, sniping wouldn't be expected to save much money with commonplace, easily valued items (as many have said). And as you noted, there are any number of flaws in the methodology of the study. And most importantly, I have made arguments above for why sniping cannot ever result in a higher final price to the winning bidder than had he/she bid the same amount at the beginning of the auction, but might result in a lower final price. Those arguments have not been refuted, although there has been a lot of babble about psychology, etc.
@CoinJunkie said:
The conclusion of the article is not that "sniping doesn't work":
In one auction of each pair we submitted an early bid, and in the other we submitted the same bid exactly ten seconds before its close. Our results, from a set of 70 such pairs, indicate no benefit to sniping: we found evidence of 2.54% lower prices for the sniped auctions, but we did not find this benefit to be statistically significant.
In fact, sniping results in slightly lower prices, which is a benefit in spite of what the authors state above. From my perspective, the only downside to sniping is that in a multiway (eBay style) auction, if you are the last person to enter a bid, you can potentially not win the auction even though your bid was the highest. That would be due to your bid not being at least a full increment higher than the current price at the time it was entered.
This is a completely ignorant statement. When the difference is not statistically significant, it means that there is effectively no difference as it is within the margin of error of the measurment.
Use common sense! There is no possible way that sniping can cost you money, but it can save you money. Them's the facts. I don't care about "statistical significance". As I stated, the downside of sniping is that you will very occasionally fail to secure a lot when you in fact submitted the highest bid, so that's the only risk in it.
That's not true. If I submit a nuclear snipe bid to try to win it - which people do - then you can end up with a higher bid than you would normally submit. You actually can't afford to be conservative, because you'll have no chance to counter.
If I submit the same bid as I normally would, no matter what, you are correct. But then I'm less likely to win because the first one to the winning number wins.
Of course, it's not my study, so I don't feel too much passion to defend the conclusion. There are other studies that show slightly different outcomes. None of them seems to show a significant price difference or a significant difference in the likelihood of winning with the sole exception [of studies I found] of a study that showed that if you found unpopular auctions with few bidders you tended to do better. Although, in that case, is the success due to sniping or simple lack of competition?
Nonetheless, I find it interesting. You may not. So be it. But please argue with data not conjecture. Thank you.
People also submit nuclear bids early in auctions. I've seen it happen many times. Just as I've seen many snipe bids of the non-nuclear variety. If you want to turn this into a debate on whether snipe bidders are likely to bid more aggressively than non-snipers, I have no interest. From the paper you cited:
In one auction of each pair we submitted an early bid, and in the other we submitted the same bid exactly ten seconds before its close.
So I believe it was reasonable to assume that that was the scenario under discussion.
And for the record, I am in no way "anti-science".
Your words: ** I don't care about "statistical significance". **
Even if they submit the identical bid early and late, since they don't always win at their maximum bid, it is still possible that the snipe bids ended at a higher number. That, of course, is NOT what the data says. The data, however, also indicates that it didn't end at a lower number.
OK, let me rephrase what I assumed was obvious: I don't care about statistical significance in the context of this argument. I'm not going to rehash everything I've already written. But to summarize: Sniping (vs. making the same bid earlier) can never cost you money, but it can save you money. No statistics are required to prove that argument.
Again, that is only true relative to your max bid. You can't ever pay more than your max bid, sure. But you could make the same case about bidding early. Bidding first can never cost you money, but it could save you money.
>
It is a completely different situation if you are talking about the price paid rather than the max bid.
If bidding $100 on day one leads to an average purchase price of $80 and sniping at $100 leads to an average purchase price of $90 - assuming the type of bid is the ONLY difference - then sniping costs you $10.
Your point is only correct in the limited and moot context of the max bid. No one EVER pays more than their max bid, no matter when placed.
You are assuming that everyone's bid, at any time, is not influenced by what's already happened. You are assuming that everyone is always going to have the same max bid. That may not be true, including the sniper. If you calculate your snipe bid based on the price on day 6 rather than the opening bid on day 1, it may end up higher. There's a whole psychology to this.
There are so many logical fallacies and misdirections in this post, I'm not sure where to begin.
For the purposes of this argument, we are talking about the methodology in the cited study. Namely, the same pre-determined maximum bid is either placed early or at the last second. There is no psychology or emotion involved there. Agree or disagree?
If you want to stick to the methodology in this study, fine. Although your statement seemed more general. But your point is still moot. As I pointed out. If you bid $100 on day one and your win price is $80 and you bid THE SAME $100 on the final day and your win price is $90. And the only difference is the timing of the bid, then bidding late cost you $10. It IS POSSIBLE, contrary to your assertion, for snipe bidding to cost you money if you are looking at price paid because...
You can NOT ignore the psychology of the other bidders even if there is no psychology to your identical $100 bids. As you mention here...
Now consider the psychology of the other bidders in the auction. If you know anything about poker and game theory, then you know that the less your opponent knows about your intentions, the more advantageous for you. By bidding at the last possible moment, you deprive the opponent(s) of the time to react to your bid. By bidding at the last second, you also eliminate the possibility of shills probing your early bid with incremental raises.
Your "opponent" does not know what your max bid is if you bid on day one. Yes, you eliminate the possibility of shills probing your early bid, assuming that's even common. But you might also consider that you change the whole nature of the auction by eliminating weaker hands early. Overall activity is, according to another study, a factor in where snipers tend to snipe.
You also need to consider the increments. Let's say your "opponent" will bid no higher than $80. If you bid early and get to $80 first, he does not place another bid. On the other hand, if he places his $80 bid first, your $100 snipe forces you to pay $82 or $85 whatever the bid increment is. That is a very simple way that the snipe would cost you MORE even with identical behavior from all bidders.
None of this is an argument against the findings of the study, which are what they are.
Why don't you just admit that you either misspoke or were incorrect? First, you suggested that the statistically insignificant difference was somehow significant. When that didn't work, you suggested that there is no way you ever pay more with a snipe than an early bid, which I've also demonstrated is simply untrue.
It's okay, we still love you even if you're wrong.
I did not misspeak. You keep coming up with off the wall scenarios to try and save face. For example, if I bid my max early, I do not force out weaker hands unless someone else has already entered a bid higher than what they (the weaker hands) wanted to bid. In which case, the weaker hands were never going to be a factor in the auction anyway.
The only point of yours I agree with is that if you bid the same exact amount as someone else, it is advantageous to have done so first. That is probably why experienced snipers tend to add a small random amount to their max bid rather than use a round number.
None of that is off the wall.
Why do i have to save face? It's not even my study you are arguing with? And you just admitted that you could end up paying more with a snipe due to the increments. So, if nothing else you are wrong on that. And, frankly, the statistics speak for themselves.
But, that's okay, I don't ever expect you to admit that you were wrong...especially if it simultaneously means I was right.
The forum may judge. I am done.
You composed the title ("bid sniping does not work"). Yet the data disclosed in the study showed that snipers saved money on average. Then the authors conclude that the savings weren't statistically significant. OK, so don't publish the damn article until you've run enough trials to achieve statistical significance! By the same token, they certainly didn't prove that sniping costs the bidder money. At worst, it's inconclusive.
I did not admit that you'll end up paying more with a snipe. You'll end up winning slightly less often, but it is impossible to construct a scenario where given the same max bids placed by the other bidders you end up paying more with a snipe than had you bid the same amount first. I dare you to try.
NO, the conclusion of the study was quite the opposite: there were NO STATISTICALLY SIGNIFICANT SAVINGS.
As for delaying publication until you have a statistically significant difference ASSUMES that they would eventually find one. Even if they run 1000 trials, there will still be a statistical test of significance. While the confidence interval will be more narrow, it is never zero. And, if there is no statistically significant difference, there is no statistically significant difference.
He's correct here, @CoinJunkie. In statistics, the lack of a statistically significant result is, itself, significant. What it's saying is that although the numbers are different (sniping saves you money) given the parameters of the trial, the difference is within the error band, and the result could have just as easily said sniping costs you money. This is the same reason that medical trials may say a drug is no better than a placebo even when more patients who received the drug get better. If not enough more get better, the results may be due to randomness (the patients who got the drug were slightly less sick than those who got the placebo) rather than because the drug actually worked.
I agree, in the abstract. However, sniping wouldn't be expected to save much money with commonplace, easily valued items (as many have said). And as you noted, there are any number of flaws in the methodology of the study. And most importantly, I have made arguments above for why sniping cannot ever result in a higher final price to the winning bidder than had he/she bid the same amount at the beginning of the auction, but might result in a lower final price. Those arguments have not been refuted, although there has been a lot of babble about psychology, etc.
And that I agree with. But the two points aren't mutually exclusive.
@CoinJunkie said:
The conclusion of the article is not that "sniping doesn't work":
In one auction of each pair we submitted an early bid, and in the other we submitted the same bid exactly ten seconds before its close. Our results, from a set of 70 such pairs, indicate no benefit to sniping: we found evidence of 2.54% lower prices for the sniped auctions, but we did not find this benefit to be statistically significant.
In fact, sniping results in slightly lower prices, which is a benefit in spite of what the authors state above. From my perspective, the only downside to sniping is that in a multiway (eBay style) auction, if you are the last person to enter a bid, you can potentially not win the auction even though your bid was the highest. That would be due to your bid not being at least a full increment higher than the current price at the time it was entered.
This is a completely ignorant statement. When the difference is not statistically significant, it means that there is effectively no difference as it is within the margin of error of the measurment.
Use common sense! There is no possible way that sniping can cost you money, but it can save you money. Them's the facts. I don't care about "statistical significance". As I stated, the downside of sniping is that you will very occasionally fail to secure a lot when you in fact submitted the highest bid, so that's the only risk in it.
That's not true. If I submit a nuclear snipe bid to try to win it - which people do - then you can end up with a higher bid than you would normally submit. You actually can't afford to be conservative, because you'll have no chance to counter.
If I submit the same bid as I normally would, no matter what, you are correct. But then I'm less likely to win because the first one to the winning number wins.
Of course, it's not my study, so I don't feel too much passion to defend the conclusion. There are other studies that show slightly different outcomes. None of them seems to show a significant price difference or a significant difference in the likelihood of winning with the sole exception [of studies I found] of a study that showed that if you found unpopular auctions with few bidders you tended to do better. Although, in that case, is the success due to sniping or simple lack of competition?
Nonetheless, I find it interesting. You may not. So be it. But please argue with data not conjecture. Thank you.
People also submit nuclear bids early in auctions. I've seen it happen many times. Just as I've seen many snipe bids of the non-nuclear variety. If you want to turn this into a debate on whether snipe bidders are likely to bid more aggressively than non-snipers, I have no interest. From the paper you cited:
In one auction of each pair we submitted an early bid, and in the other we submitted the same bid exactly ten seconds before its close.
So I believe it was reasonable to assume that that was the scenario under discussion.
And for the record, I am in no way "anti-science".
Your words: ** I don't care about "statistical significance". **
Even if they submit the identical bid early and late, since they don't always win at their maximum bid, it is still possible that the snipe bids ended at a higher number. That, of course, is NOT what the data says. The data, however, also indicates that it didn't end at a lower number.
OK, let me rephrase what I assumed was obvious: I don't care about statistical significance in the context of this argument. I'm not going to rehash everything I've already written. But to summarize: Sniping (vs. making the same bid earlier) can never cost you money, but it can save you money. No statistics are required to prove that argument.
Again, that is only true relative to your max bid. You can't ever pay more than your max bid, sure. But you could make the same case about bidding early. Bidding first can never cost you money, but it could save you money.
>
It is a completely different situation if you are talking about the price paid rather than the max bid.
If bidding $100 on day one leads to an average purchase price of $80 and sniping at $100 leads to an average purchase price of $90 - assuming the type of bid is the ONLY difference - then sniping costs you $10.
Your point is only correct in the limited and moot context of the max bid. No one EVER pays more than their max bid, no matter when placed.
You are assuming that everyone's bid, at any time, is not influenced by what's already happened. You are assuming that everyone is always going to have the same max bid. That may not be true, including the sniper. If you calculate your snipe bid based on the price on day 6 rather than the opening bid on day 1, it may end up higher. There's a whole psychology to this.
There are so many logical fallacies and misdirections in this post, I'm not sure where to begin.
For the purposes of this argument, we are talking about the methodology in the cited study. Namely, the same pre-determined maximum bid is either placed early or at the last second. There is no psychology or emotion involved there. Agree or disagree?
If you want to stick to the methodology in this study, fine. Although your statement seemed more general. But your point is still moot. As I pointed out. If you bid $100 on day one and your win price is $80 and you bid THE SAME $100 on the final day and your win price is $90. And the only difference is the timing of the bid, then bidding late cost you $10. It IS POSSIBLE, contrary to your assertion, for snipe bidding to cost you money if you are looking at price paid because...
You can NOT ignore the psychology of the other bidders even if there is no psychology to your identical $100 bids. As you mention here...
Now consider the psychology of the other bidders in the auction. If you know anything about poker and game theory, then you know that the less your opponent knows about your intentions, the more advantageous for you. By bidding at the last possible moment, you deprive the opponent(s) of the time to react to your bid. By bidding at the last second, you also eliminate the possibility of shills probing your early bid with incremental raises.
Your "opponent" does not know what your max bid is if you bid on day one. Yes, you eliminate the possibility of shills probing your early bid, assuming that's even common. But you might also consider that you change the whole nature of the auction by eliminating weaker hands early. Overall activity is, according to another study, a factor in where snipers tend to snipe.
You also need to consider the increments. Let's say your "opponent" will bid no higher than $80. If you bid early and get to $80 first, he does not place another bid. On the other hand, if he places his $80 bid first, your $100 snipe forces you to pay $82 or $85 whatever the bid increment is. That is a very simple way that the snipe would cost you MORE even with identical behavior from all bidders.
None of this is an argument against the findings of the study, which are what they are.
Why don't you just admit that you either misspoke or were incorrect? First, you suggested that the statistically insignificant difference was somehow significant. When that didn't work, you suggested that there is no way you ever pay more with a snipe than an early bid, which I've also demonstrated is simply untrue.
It's okay, we still love you even if you're wrong.
I did not misspeak. You keep coming up with off the wall scenarios to try and save face. For example, if I bid my max early, I do not force out weaker hands unless someone else has already entered a bid higher than what they (the weaker hands) wanted to bid. In which case, the weaker hands were never going to be a factor in the auction anyway.
The only point of yours I agree with is that if you bid the same exact amount as someone else, it is advantageous to have done so first. That is probably why experienced snipers tend to add a small random amount to their max bid rather than use a round number.
None of that is off the wall.
Why do i have to save face? It's not even my study you are arguing with? And you just admitted that you could end up paying more with a snipe due to the increments. So, if nothing else you are wrong on that. And, frankly, the statistics speak for themselves.
But, that's okay, I don't ever expect you to admit that you were wrong...especially if it simultaneously means I was right.
The forum may judge. I am done.
You composed the title ("bid sniping does not work"). Yet the data disclosed in the study showed that snipers saved money on average. Then the authors conclude that the savings weren't statistically significant. OK, so don't publish the damn article until you've run enough trials to achieve statistical significance! By the same token, they certainly didn't prove that sniping costs the bidder money. At worst, it's inconclusive.
I did not admit that you'll end up paying more with a snipe. You'll end up winning slightly less often, but it is impossible to construct a scenario where given the same max bids placed by the other bidders you end up paying more with a snipe than had you bid the same amount first. I dare you to try.
NO, the conclusion of the study was quite the opposite: there were NO STATISTICALLY SIGNIFICANT SAVINGS.
As for delaying publication until you have a statistically significant difference ASSUMES that they would eventually find one. Even if they run 1000 trials, there will still be a statistical test of significance. While the confidence interval will be more narrow, it is never zero. And, if there is no statistically significant difference, there is no statistically significant difference.
He's correct here, @CoinJunkie. In statistics, the lack of a statistically significant result is, itself, significant. What it's saying is that although the numbers are different (sniping saves you money) given the parameters of the trial, the difference is within the error band, and the result could have just as easily said sniping costs you money. This is the same reason that medical trials may say a drug is no better than a placebo even when more patients who received the drug get better. If not enough more get better, the results may be due to randomness (the patients who got the drug were slightly less sick than those who got the placebo) rather than because the drug actually worked.
I agree, in the abstract. However, sniping wouldn't be expected to save much money with commonplace, easily valued items (as many have said). And as you noted, there are any number of flaws in the methodology of the study. And most importantly, I have made arguments above for why sniping cannot ever result in a higher final price to the winning bidder than had he/she bid the same amount at the beginning of the auction, but might result in a lower final price. Those arguments have not been refuted, although there has been a lot of babble about psychology, etc.
ACTUALLY THEY HAVE BEEN REFUTED. You just refuse to acknowledge it.
Bit $100 on day 1. Underbidder won't go past $80. You are the first one to $80, you win.
You wait until the last day. Underbidder has already bid $80. You have to go one increment higher.
I can't make it any simpler than that. Even if EVERYONE bids exactly the same - no psychology or anything - you CAN pay more by sniping.
All comments reflect the opinion of the author, even when irrefutably accurate.
@CoinJunkie said:
The conclusion of the article is not that "sniping doesn't work":
In one auction of each pair we submitted an early bid, and in the other we submitted the same bid exactly ten seconds before its close. Our results, from a set of 70 such pairs, indicate no benefit to sniping: we found evidence of 2.54% lower prices for the sniped auctions, but we did not find this benefit to be statistically significant.
In fact, sniping results in slightly lower prices, which is a benefit in spite of what the authors state above. From my perspective, the only downside to sniping is that in a multiway (eBay style) auction, if you are the last person to enter a bid, you can potentially not win the auction even though your bid was the highest. That would be due to your bid not being at least a full increment higher than the current price at the time it was entered.
This is a completely ignorant statement. When the difference is not statistically significant, it means that there is effectively no difference as it is within the margin of error of the measurment.
Use common sense! There is no possible way that sniping can cost you money, but it can save you money. Them's the facts. I don't care about "statistical significance". As I stated, the downside of sniping is that you will very occasionally fail to secure a lot when you in fact submitted the highest bid, so that's the only risk in it.
That's not true. If I submit a nuclear snipe bid to try to win it - which people do - then you can end up with a higher bid than you would normally submit. You actually can't afford to be conservative, because you'll have no chance to counter.
If I submit the same bid as I normally would, no matter what, you are correct. But then I'm less likely to win because the first one to the winning number wins.
Of course, it's not my study, so I don't feel too much passion to defend the conclusion. There are other studies that show slightly different outcomes. None of them seems to show a significant price difference or a significant difference in the likelihood of winning with the sole exception [of studies I found] of a study that showed that if you found unpopular auctions with few bidders you tended to do better. Although, in that case, is the success due to sniping or simple lack of competition?
Nonetheless, I find it interesting. You may not. So be it. But please argue with data not conjecture. Thank you.
People also submit nuclear bids early in auctions. I've seen it happen many times. Just as I've seen many snipe bids of the non-nuclear variety. If you want to turn this into a debate on whether snipe bidders are likely to bid more aggressively than non-snipers, I have no interest. From the paper you cited:
In one auction of each pair we submitted an early bid, and in the other we submitted the same bid exactly ten seconds before its close.
So I believe it was reasonable to assume that that was the scenario under discussion.
And for the record, I am in no way "anti-science".
Your words: ** I don't care about "statistical significance". **
Even if they submit the identical bid early and late, since they don't always win at their maximum bid, it is still possible that the snipe bids ended at a higher number. That, of course, is NOT what the data says. The data, however, also indicates that it didn't end at a lower number.
OK, let me rephrase what I assumed was obvious: I don't care about statistical significance in the context of this argument. I'm not going to rehash everything I've already written. But to summarize: Sniping (vs. making the same bid earlier) can never cost you money, but it can save you money. No statistics are required to prove that argument.
Again, that is only true relative to your max bid. You can't ever pay more than your max bid, sure. But you could make the same case about bidding early. Bidding first can never cost you money, but it could save you money.
>
It is a completely different situation if you are talking about the price paid rather than the max bid.
If bidding $100 on day one leads to an average purchase price of $80 and sniping at $100 leads to an average purchase price of $90 - assuming the type of bid is the ONLY difference - then sniping costs you $10.
Your point is only correct in the limited and moot context of the max bid. No one EVER pays more than their max bid, no matter when placed.
You are assuming that everyone's bid, at any time, is not influenced by what's already happened. You are assuming that everyone is always going to have the same max bid. That may not be true, including the sniper. If you calculate your snipe bid based on the price on day 6 rather than the opening bid on day 1, it may end up higher. There's a whole psychology to this.
There are so many logical fallacies and misdirections in this post, I'm not sure where to begin.
For the purposes of this argument, we are talking about the methodology in the cited study. Namely, the same pre-determined maximum bid is either placed early or at the last second. There is no psychology or emotion involved there. Agree or disagree?
If you want to stick to the methodology in this study, fine. Although your statement seemed more general. But your point is still moot. As I pointed out. If you bid $100 on day one and your win price is $80 and you bid THE SAME $100 on the final day and your win price is $90. And the only difference is the timing of the bid, then bidding late cost you $10. It IS POSSIBLE, contrary to your assertion, for snipe bidding to cost you money if you are looking at price paid because...
You can NOT ignore the psychology of the other bidders even if there is no psychology to your identical $100 bids. As you mention here...
Now consider the psychology of the other bidders in the auction. If you know anything about poker and game theory, then you know that the less your opponent knows about your intentions, the more advantageous for you. By bidding at the last possible moment, you deprive the opponent(s) of the time to react to your bid. By bidding at the last second, you also eliminate the possibility of shills probing your early bid with incremental raises.
Your "opponent" does not know what your max bid is if you bid on day one. Yes, you eliminate the possibility of shills probing your early bid, assuming that's even common. But you might also consider that you change the whole nature of the auction by eliminating weaker hands early. Overall activity is, according to another study, a factor in where snipers tend to snipe.
You also need to consider the increments. Let's say your "opponent" will bid no higher than $80. If you bid early and get to $80 first, he does not place another bid. On the other hand, if he places his $80 bid first, your $100 snipe forces you to pay $82 or $85 whatever the bid increment is. That is a very simple way that the snipe would cost you MORE even with identical behavior from all bidders.
None of this is an argument against the findings of the study, which are what they are.
Why don't you just admit that you either misspoke or were incorrect? First, you suggested that the statistically insignificant difference was somehow significant. When that didn't work, you suggested that there is no way you ever pay more with a snipe than an early bid, which I've also demonstrated is simply untrue.
It's okay, we still love you even if you're wrong.
I did not misspeak. You keep coming up with off the wall scenarios to try and save face. For example, if I bid my max early, I do not force out weaker hands unless someone else has already entered a bid higher than what they (the weaker hands) wanted to bid. In which case, the weaker hands were never going to be a factor in the auction anyway.
The only point of yours I agree with is that if you bid the same exact amount as someone else, it is advantageous to have done so first. That is probably why experienced snipers tend to add a small random amount to their max bid rather than use a round number.
None of that is off the wall.
Why do i have to save face? It's not even my study you are arguing with? And you just admitted that you could end up paying more with a snipe due to the increments. So, if nothing else you are wrong on that. And, frankly, the statistics speak for themselves.
But, that's okay, I don't ever expect you to admit that you were wrong...especially if it simultaneously means I was right.
The forum may judge. I am done.
You composed the title ("bid sniping does not work"). Yet the data disclosed in the study showed that snipers saved money on average. Then the authors conclude that the savings weren't statistically significant. OK, so don't publish the damn article until you've run enough trials to achieve statistical significance! By the same token, they certainly didn't prove that sniping costs the bidder money. At worst, it's inconclusive.
I did not admit that you'll end up paying more with a snipe. You'll end up winning slightly less often, but it is impossible to construct a scenario where given the same max bids placed by the other bidders you end up paying more with a snipe than had you bid the same amount first. I dare you to try.
NO, the conclusion of the study was quite the opposite: there were NO STATISTICALLY SIGNIFICANT SAVINGS.
As for delaying publication until you have a statistically significant difference ASSUMES that they would eventually find one. Even if they run 1000 trials, there will still be a statistical test of significance. While the confidence interval will be more narrow, it is never zero. And, if there is no statistically significant difference, there is no statistically significant difference.
He's correct here, @CoinJunkie. In statistics, the lack of a statistically significant result is, itself, significant. What it's saying is that although the numbers are different (sniping saves you money) given the parameters of the trial, the difference is within the error band, and the result could have just as easily said sniping costs you money. This is the same reason that medical trials may say a drug is no better than a placebo even when more patients who received the drug get better. If not enough more get better, the results may be due to randomness (the patients who got the drug were slightly less sick than those who got the placebo) rather than because the drug actually worked.
I agree, in the abstract. However, sniping wouldn't be expected to save much money with commonplace, easily valued items (as many have said). And as you noted, there are any number of flaws in the methodology of the study. And most importantly, I have made arguments above for why sniping cannot ever result in a higher final price to the winning bidder than had he/she bid the same amount at the beginning of the auction, but might result in a lower final price. Those arguments have not been refuted, although there has been a lot of babble about psychology, etc.
ACTUALLY THEY HAVE BEEN REFUTED. You just refuse to acknowledge it.
Bit $100 on day 1. Underbidder won't go past $80. You are the first one to $80, you win.
You wait until the last day. Underbidder has already bid $80. You have to go one increment higher.
I can't make it any simpler than that. Even if EVERYONE bids exactly the same - no psychology or anything - you CAN pay more by sniping.
READ the conditions of the thought experiment!!! You make the same max bid either at the very beginning or at the very end of the auction. In the example you present, when you snipe, you LOSE the auction, you do not "have to go one increment higher". I've already pointed out multiple times that the downside to sniping is that you will occasionally lose when your max bid is equal to or higher than any other bidder's.
The logical conclusion is that if you prioritize winning the lot over getting the best possible price, sniping may not (although that is arguable) be a good choice. If getting the best possible price on any lot won is the priority, then sniping is DEMONSTRABLY (without running ANY simulations or empirical studies) your best strategy. The savings probably won't make you rich, but they are savings nonetheless.
@CoinJunkie said:
The conclusion of the article is not that "sniping doesn't work":
In one auction of each pair we submitted an early bid, and in the other we submitted the same bid exactly ten seconds before its close. Our results, from a set of 70 such pairs, indicate no benefit to sniping: we found evidence of 2.54% lower prices for the sniped auctions, but we did not find this benefit to be statistically significant.
In fact, sniping results in slightly lower prices, which is a benefit in spite of what the authors state above. From my perspective, the only downside to sniping is that in a multiway (eBay style) auction, if you are the last person to enter a bid, you can potentially not win the auction even though your bid was the highest. That would be due to your bid not being at least a full increment higher than the current price at the time it was entered.
This is a completely ignorant statement. When the difference is not statistically significant, it means that there is effectively no difference as it is within the margin of error of the measurment.
Use common sense! There is no possible way that sniping can cost you money, but it can save you money. Them's the facts. I don't care about "statistical significance". As I stated, the downside of sniping is that you will very occasionally fail to secure a lot when you in fact submitted the highest bid, so that's the only risk in it.
That's not true. If I submit a nuclear snipe bid to try to win it - which people do - then you can end up with a higher bid than you would normally submit. You actually can't afford to be conservative, because you'll have no chance to counter.
If I submit the same bid as I normally would, no matter what, you are correct. But then I'm less likely to win because the first one to the winning number wins.
Of course, it's not my study, so I don't feel too much passion to defend the conclusion. There are other studies that show slightly different outcomes. None of them seems to show a significant price difference or a significant difference in the likelihood of winning with the sole exception [of studies I found] of a study that showed that if you found unpopular auctions with few bidders you tended to do better. Although, in that case, is the success due to sniping or simple lack of competition?
Nonetheless, I find it interesting. You may not. So be it. But please argue with data not conjecture. Thank you.
People also submit nuclear bids early in auctions. I've seen it happen many times. Just as I've seen many snipe bids of the non-nuclear variety. If you want to turn this into a debate on whether snipe bidders are likely to bid more aggressively than non-snipers, I have no interest. From the paper you cited:
In one auction of each pair we submitted an early bid, and in the other we submitted the same bid exactly ten seconds before its close.
So I believe it was reasonable to assume that that was the scenario under discussion.
And for the record, I am in no way "anti-science".
Your words: ** I don't care about "statistical significance". **
Even if they submit the identical bid early and late, since they don't always win at their maximum bid, it is still possible that the snipe bids ended at a higher number. That, of course, is NOT what the data says. The data, however, also indicates that it didn't end at a lower number.
OK, let me rephrase what I assumed was obvious: I don't care about statistical significance in the context of this argument. I'm not going to rehash everything I've already written. But to summarize: Sniping (vs. making the same bid earlier) can never cost you money, but it can save you money. No statistics are required to prove that argument.
Again, that is only true relative to your max bid. You can't ever pay more than your max bid, sure. But you could make the same case about bidding early. Bidding first can never cost you money, but it could save you money.
>
It is a completely different situation if you are talking about the price paid rather than the max bid.
If bidding $100 on day one leads to an average purchase price of $80 and sniping at $100 leads to an average purchase price of $90 - assuming the type of bid is the ONLY difference - then sniping costs you $10.
Your point is only correct in the limited and moot context of the max bid. No one EVER pays more than their max bid, no matter when placed.
You are assuming that everyone's bid, at any time, is not influenced by what's already happened. You are assuming that everyone is always going to have the same max bid. That may not be true, including the sniper. If you calculate your snipe bid based on the price on day 6 rather than the opening bid on day 1, it may end up higher. There's a whole psychology to this.
There are so many logical fallacies and misdirections in this post, I'm not sure where to begin.
For the purposes of this argument, we are talking about the methodology in the cited study. Namely, the same pre-determined maximum bid is either placed early or at the last second. There is no psychology or emotion involved there. Agree or disagree?
If you want to stick to the methodology in this study, fine. Although your statement seemed more general. But your point is still moot. As I pointed out. If you bid $100 on day one and your win price is $80 and you bid THE SAME $100 on the final day and your win price is $90. And the only difference is the timing of the bid, then bidding late cost you $10. It IS POSSIBLE, contrary to your assertion, for snipe bidding to cost you money if you are looking at price paid because...
You can NOT ignore the psychology of the other bidders even if there is no psychology to your identical $100 bids. As you mention here...
Now consider the psychology of the other bidders in the auction. If you know anything about poker and game theory, then you know that the less your opponent knows about your intentions, the more advantageous for you. By bidding at the last possible moment, you deprive the opponent(s) of the time to react to your bid. By bidding at the last second, you also eliminate the possibility of shills probing your early bid with incremental raises.
Your "opponent" does not know what your max bid is if you bid on day one. Yes, you eliminate the possibility of shills probing your early bid, assuming that's even common. But you might also consider that you change the whole nature of the auction by eliminating weaker hands early. Overall activity is, according to another study, a factor in where snipers tend to snipe.
You also need to consider the increments. Let's say your "opponent" will bid no higher than $80. If you bid early and get to $80 first, he does not place another bid. On the other hand, if he places his $80 bid first, your $100 snipe forces you to pay $82 or $85 whatever the bid increment is. That is a very simple way that the snipe would cost you MORE even with identical behavior from all bidders.
None of this is an argument against the findings of the study, which are what they are.
Why don't you just admit that you either misspoke or were incorrect? First, you suggested that the statistically insignificant difference was somehow significant. When that didn't work, you suggested that there is no way you ever pay more with a snipe than an early bid, which I've also demonstrated is simply untrue.
It's okay, we still love you even if you're wrong.
I did not misspeak. You keep coming up with off the wall scenarios to try and save face. For example, if I bid my max early, I do not force out weaker hands unless someone else has already entered a bid higher than what they (the weaker hands) wanted to bid. In which case, the weaker hands were never going to be a factor in the auction anyway.
The only point of yours I agree with is that if you bid the same exact amount as someone else, it is advantageous to have done so first. That is probably why experienced snipers tend to add a small random amount to their max bid rather than use a round number.
None of that is off the wall.
Why do i have to save face? It's not even my study you are arguing with? And you just admitted that you could end up paying more with a snipe due to the increments. So, if nothing else you are wrong on that. And, frankly, the statistics speak for themselves.
But, that's okay, I don't ever expect you to admit that you were wrong...especially if it simultaneously means I was right.
The forum may judge. I am done.
You composed the title ("bid sniping does not work"). Yet the data disclosed in the study showed that snipers saved money on average. Then the authors conclude that the savings weren't statistically significant. OK, so don't publish the damn article until you've run enough trials to achieve statistical significance! By the same token, they certainly didn't prove that sniping costs the bidder money. At worst, it's inconclusive.
I did not admit that you'll end up paying more with a snipe. You'll end up winning slightly less often, but it is impossible to construct a scenario where given the same max bids placed by the other bidders you end up paying more with a snipe than had you bid the same amount first. I dare you to try.
NO, the conclusion of the study was quite the opposite: there were NO STATISTICALLY SIGNIFICANT SAVINGS.
As for delaying publication until you have a statistically significant difference ASSUMES that they would eventually find one. Even if they run 1000 trials, there will still be a statistical test of significance. While the confidence interval will be more narrow, it is never zero. And, if there is no statistically significant difference, there is no statistically significant difference.
He's correct here, @CoinJunkie. In statistics, the lack of a statistically significant result is, itself, significant. What it's saying is that although the numbers are different (sniping saves you money) given the parameters of the trial, the difference is within the error band, and the result could have just as easily said sniping costs you money. This is the same reason that medical trials may say a drug is no better than a placebo even when more patients who received the drug get better. If not enough more get better, the results may be due to randomness (the patients who got the drug were slightly less sick than those who got the placebo) rather than because the drug actually worked.
I agree, in the abstract. However, sniping wouldn't be expected to save much money with commonplace, easily valued items (as many have said). And as you noted, there are any number of flaws in the methodology of the study. And most importantly, I have made arguments above for why sniping cannot ever result in a higher final price to the winning bidder than had he/she bid the same amount at the beginning of the auction, but might result in a lower final price. Those arguments have not been refuted, although there has been a lot of babble about psychology, etc.
ACTUALLY THEY HAVE BEEN REFUTED. You just refuse to acknowledge it.
Bit $100 on day 1. Underbidder won't go past $80. You are the first one to $80, you win.
You wait until the last day. Underbidder has already bid $80. You have to go one increment higher.
I can't make it any simpler than that. Even if EVERYONE bids exactly the same - no psychology or anything - you CAN pay more by sniping.
READ the conditions of the thought experiment!!! You make the same max bid either at the very beginning or at the very end of the auction. In the example you present, when you snipe, you LOSE the auction, you do not "have to go one increment higher". I've already pointed out multiple times that the downside to sniping is that you will occasionally lose when your max bid is equal to or higher than any other bidder's.
The logical conclusion is that if you prioritize winning the lot over getting the best possible price, sniping may not (although that is arguable) be a good choice. If getting the best possible price on any lot won is the priority, then sniping is DEMONSTRABLY (without running ANY simulations or empirical studies) your best strategy. The savings probably won't make you rich, but they are savings nonetheless.
Over and out!
That is not true. Dear God, read slower.
You bid $100 in both cases. You WIN In both cases. It costs you $80 when you bid day 1. It costs you $82 when you sniped.
CONTRARY to your claim that sniping can never cost you more.
Q.E.D.
All comments reflect the opinion of the author, even when irrefutably accurate.
I think the study is valid - but only for what they tested.
THEY TESTED WIDGETS! Items a population of a bazillion and that most people (including me) that wouldn't waste their time on a snipe bid because surely another widget will come along.
It is not a surprise that they came to the conclusion that they did.
Rerun the test with items that deserve sniping and I suspect you will get a far different result.
“In matters of style, swim with the current; in matters of principle, stand like a rock." - Thomas Jefferson
@Cameonut said:
I think the study is valid - but only for what they tested.
THEY TESTED WIDGETS! Items a population of a bazillion and that most people (including me) that wouldn't waste their time on a snipe bid because surely another widget will come along.
It is not a surprise that they came to the conclusion that they did.
Rerun the test with items that deserve sniping and I suspect you will get a far different result.
It's hard to run the test with non-widgets. If you can't pair the "identical" coins, how do you know whether the resulting price difference is due to the timing of the bid or the coin itself?
I think the bigger problem with sniping on "unique" coins is that you have the possibility of two people going nuclear. It must have happened here or there, don't you think?
All comments reflect the opinion of the author, even when irrefutably accurate.
@CoinJunkie said:
The conclusion of the article is not that "sniping doesn't work":
In one auction of each pair we submitted an early bid, and in the other we submitted the same bid exactly ten seconds before its close. Our results, from a set of 70 such pairs, indicate no benefit to sniping: we found evidence of 2.54% lower prices for the sniped auctions, but we did not find this benefit to be statistically significant.
In fact, sniping results in slightly lower prices, which is a benefit in spite of what the authors state above. From my perspective, the only downside to sniping is that in a multiway (eBay style) auction, if you are the last person to enter a bid, you can potentially not win the auction even though your bid was the highest. That would be due to your bid not being at least a full increment higher than the current price at the time it was entered.
This is a completely ignorant statement. When the difference is not statistically significant, it means that there is effectively no difference as it is within the margin of error of the measurment.
Use common sense! There is no possible way that sniping can cost you money, but it can save you money. Them's the facts. I don't care about "statistical significance". As I stated, the downside of sniping is that you will very occasionally fail to secure a lot when you in fact submitted the highest bid, so that's the only risk in it.
That's not true. If I submit a nuclear snipe bid to try to win it - which people do - then you can end up with a higher bid than you would normally submit. You actually can't afford to be conservative, because you'll have no chance to counter.
If I submit the same bid as I normally would, no matter what, you are correct. But then I'm less likely to win because the first one to the winning number wins.
Of course, it's not my study, so I don't feel too much passion to defend the conclusion. There are other studies that show slightly different outcomes. None of them seems to show a significant price difference or a significant difference in the likelihood of winning with the sole exception [of studies I found] of a study that showed that if you found unpopular auctions with few bidders you tended to do better. Although, in that case, is the success due to sniping or simple lack of competition?
Nonetheless, I find it interesting. You may not. So be it. But please argue with data not conjecture. Thank you.
People also submit nuclear bids early in auctions. I've seen it happen many times. Just as I've seen many snipe bids of the non-nuclear variety. If you want to turn this into a debate on whether snipe bidders are likely to bid more aggressively than non-snipers, I have no interest. From the paper you cited:
In one auction of each pair we submitted an early bid, and in the other we submitted the same bid exactly ten seconds before its close.
So I believe it was reasonable to assume that that was the scenario under discussion.
And for the record, I am in no way "anti-science".
Your words: ** I don't care about "statistical significance". **
Even if they submit the identical bid early and late, since they don't always win at their maximum bid, it is still possible that the snipe bids ended at a higher number. That, of course, is NOT what the data says. The data, however, also indicates that it didn't end at a lower number.
OK, let me rephrase what I assumed was obvious: I don't care about statistical significance in the context of this argument. I'm not going to rehash everything I've already written. But to summarize: Sniping (vs. making the same bid earlier) can never cost you money, but it can save you money. No statistics are required to prove that argument.
Again, that is only true relative to your max bid. You can't ever pay more than your max bid, sure. But you could make the same case about bidding early. Bidding first can never cost you money, but it could save you money.
>
It is a completely different situation if you are talking about the price paid rather than the max bid.
If bidding $100 on day one leads to an average purchase price of $80 and sniping at $100 leads to an average purchase price of $90 - assuming the type of bid is the ONLY difference - then sniping costs you $10.
Your point is only correct in the limited and moot context of the max bid. No one EVER pays more than their max bid, no matter when placed.
You are assuming that everyone's bid, at any time, is not influenced by what's already happened. You are assuming that everyone is always going to have the same max bid. That may not be true, including the sniper. If you calculate your snipe bid based on the price on day 6 rather than the opening bid on day 1, it may end up higher. There's a whole psychology to this.
There are so many logical fallacies and misdirections in this post, I'm not sure where to begin.
For the purposes of this argument, we are talking about the methodology in the cited study. Namely, the same pre-determined maximum bid is either placed early or at the last second. There is no psychology or emotion involved there. Agree or disagree?
If you want to stick to the methodology in this study, fine. Although your statement seemed more general. But your point is still moot. As I pointed out. If you bid $100 on day one and your win price is $80 and you bid THE SAME $100 on the final day and your win price is $90. And the only difference is the timing of the bid, then bidding late cost you $10. It IS POSSIBLE, contrary to your assertion, for snipe bidding to cost you money if you are looking at price paid because...
You can NOT ignore the psychology of the other bidders even if there is no psychology to your identical $100 bids. As you mention here...
Now consider the psychology of the other bidders in the auction. If you know anything about poker and game theory, then you know that the less your opponent knows about your intentions, the more advantageous for you. By bidding at the last possible moment, you deprive the opponent(s) of the time to react to your bid. By bidding at the last second, you also eliminate the possibility of shills probing your early bid with incremental raises.
Your "opponent" does not know what your max bid is if you bid on day one. Yes, you eliminate the possibility of shills probing your early bid, assuming that's even common. But you might also consider that you change the whole nature of the auction by eliminating weaker hands early. Overall activity is, according to another study, a factor in where snipers tend to snipe.
You also need to consider the increments. Let's say your "opponent" will bid no higher than $80. If you bid early and get to $80 first, he does not place another bid. On the other hand, if he places his $80 bid first, your $100 snipe forces you to pay $82 or $85 whatever the bid increment is. That is a very simple way that the snipe would cost you MORE even with identical behavior from all bidders.
None of this is an argument against the findings of the study, which are what they are.
Why don't you just admit that you either misspoke or were incorrect? First, you suggested that the statistically insignificant difference was somehow significant. When that didn't work, you suggested that there is no way you ever pay more with a snipe than an early bid, which I've also demonstrated is simply untrue.
It's okay, we still love you even if you're wrong.
I did not misspeak. You keep coming up with off the wall scenarios to try and save face. For example, if I bid my max early, I do not force out weaker hands unless someone else has already entered a bid higher than what they (the weaker hands) wanted to bid. In which case, the weaker hands were never going to be a factor in the auction anyway.
The only point of yours I agree with is that if you bid the same exact amount as someone else, it is advantageous to have done so first. That is probably why experienced snipers tend to add a small random amount to their max bid rather than use a round number.
None of that is off the wall.
Why do i have to save face? It's not even my study you are arguing with? And you just admitted that you could end up paying more with a snipe due to the increments. So, if nothing else you are wrong on that. And, frankly, the statistics speak for themselves.
But, that's okay, I don't ever expect you to admit that you were wrong...especially if it simultaneously means I was right.
The forum may judge. I am done.
You composed the title ("bid sniping does not work"). Yet the data disclosed in the study showed that snipers saved money on average. Then the authors conclude that the savings weren't statistically significant. OK, so don't publish the damn article until you've run enough trials to achieve statistical significance! By the same token, they certainly didn't prove that sniping costs the bidder money. At worst, it's inconclusive.
I did not admit that you'll end up paying more with a snipe. You'll end up winning slightly less often, but it is impossible to construct a scenario where given the same max bids placed by the other bidders you end up paying more with a snipe than had you bid the same amount first. I dare you to try.
NO, the conclusion of the study was quite the opposite: there were NO STATISTICALLY SIGNIFICANT SAVINGS.
As for delaying publication until you have a statistically significant difference ASSUMES that they would eventually find one. Even if they run 1000 trials, there will still be a statistical test of significance. While the confidence interval will be more narrow, it is never zero. And, if there is no statistically significant difference, there is no statistically significant difference.
He's correct here, @CoinJunkie. In statistics, the lack of a statistically significant result is, itself, significant. What it's saying is that although the numbers are different (sniping saves you money) given the parameters of the trial, the difference is within the error band, and the result could have just as easily said sniping costs you money. This is the same reason that medical trials may say a drug is no better than a placebo even when more patients who received the drug get better. If not enough more get better, the results may be due to randomness (the patients who got the drug were slightly less sick than those who got the placebo) rather than because the drug actually worked.
I agree, in the abstract. However, sniping wouldn't be expected to save much money with commonplace, easily valued items (as many have said). And as you noted, there are any number of flaws in the methodology of the study. And most importantly, I have made arguments above for why sniping cannot ever result in a higher final price to the winning bidder than had he/she bid the same amount at the beginning of the auction, but might result in a lower final price. Those arguments have not been refuted, although there has been a lot of babble about psychology, etc.
ACTUALLY THEY HAVE BEEN REFUTED. You just refuse to acknowledge it.
Bit $100 on day 1. Underbidder won't go past $80. You are the first one to $80, you win.
You wait until the last day. Underbidder has already bid $80. You have to go one increment higher.
I can't make it any simpler than that. Even if EVERYONE bids exactly the same - no psychology or anything - you CAN pay more by sniping.
This is incorrect. The winning bid would be $81 either way. If you bid $100 first and then the underbidder won't go past $80 and bids that amount, ebay will automatically bump your bid to $81 if you have previously bid $100 as your high bid. That is how proxy bidding works. In the latter case, the underbidder bids the same $80 and with your snipe, you outbid his $80 and win the item at the next bid increment of $81.
Collecting 1970s Topps baseball wax, rack and cello packs, as well as PCGS graded Half Cents, Large Cents, Two Cent pieces and Three Cent Silver pieces.
@CoinJunkie said:
The conclusion of the article is not that "sniping doesn't work":
In one auction of each pair we submitted an early bid, and in the other we submitted the same bid exactly ten seconds before its close. Our results, from a set of 70 such pairs, indicate no benefit to sniping: we found evidence of 2.54% lower prices for the sniped auctions, but we did not find this benefit to be statistically significant.
In fact, sniping results in slightly lower prices, which is a benefit in spite of what the authors state above. From my perspective, the only downside to sniping is that in a multiway (eBay style) auction, if you are the last person to enter a bid, you can potentially not win the auction even though your bid was the highest. That would be due to your bid not being at least a full increment higher than the current price at the time it was entered.
This is a completely ignorant statement. When the difference is not statistically significant, it means that there is effectively no difference as it is within the margin of error of the measurment.
Use common sense! There is no possible way that sniping can cost you money, but it can save you money. Them's the facts. I don't care about "statistical significance". As I stated, the downside of sniping is that you will very occasionally fail to secure a lot when you in fact submitted the highest bid, so that's the only risk in it.
That's not true. If I submit a nuclear snipe bid to try to win it - which people do - then you can end up with a higher bid than you would normally submit. You actually can't afford to be conservative, because you'll have no chance to counter.
If I submit the same bid as I normally would, no matter what, you are correct. But then I'm less likely to win because the first one to the winning number wins.
Of course, it's not my study, so I don't feel too much passion to defend the conclusion. There are other studies that show slightly different outcomes. None of them seems to show a significant price difference or a significant difference in the likelihood of winning with the sole exception [of studies I found] of a study that showed that if you found unpopular auctions with few bidders you tended to do better. Although, in that case, is the success due to sniping or simple lack of competition?
Nonetheless, I find it interesting. You may not. So be it. But please argue with data not conjecture. Thank you.
People also submit nuclear bids early in auctions. I've seen it happen many times. Just as I've seen many snipe bids of the non-nuclear variety. If you want to turn this into a debate on whether snipe bidders are likely to bid more aggressively than non-snipers, I have no interest. From the paper you cited:
In one auction of each pair we submitted an early bid, and in the other we submitted the same bid exactly ten seconds before its close.
So I believe it was reasonable to assume that that was the scenario under discussion.
And for the record, I am in no way "anti-science".
Your words: ** I don't care about "statistical significance". **
Even if they submit the identical bid early and late, since they don't always win at their maximum bid, it is still possible that the snipe bids ended at a higher number. That, of course, is NOT what the data says. The data, however, also indicates that it didn't end at a lower number.
OK, let me rephrase what I assumed was obvious: I don't care about statistical significance in the context of this argument. I'm not going to rehash everything I've already written. But to summarize: Sniping (vs. making the same bid earlier) can never cost you money, but it can save you money. No statistics are required to prove that argument.
Again, that is only true relative to your max bid. You can't ever pay more than your max bid, sure. But you could make the same case about bidding early. Bidding first can never cost you money, but it could save you money.
>
It is a completely different situation if you are talking about the price paid rather than the max bid.
If bidding $100 on day one leads to an average purchase price of $80 and sniping at $100 leads to an average purchase price of $90 - assuming the type of bid is the ONLY difference - then sniping costs you $10.
Your point is only correct in the limited and moot context of the max bid. No one EVER pays more than their max bid, no matter when placed.
You are assuming that everyone's bid, at any time, is not influenced by what's already happened. You are assuming that everyone is always going to have the same max bid. That may not be true, including the sniper. If you calculate your snipe bid based on the price on day 6 rather than the opening bid on day 1, it may end up higher. There's a whole psychology to this.
There are so many logical fallacies and misdirections in this post, I'm not sure where to begin.
For the purposes of this argument, we are talking about the methodology in the cited study. Namely, the same pre-determined maximum bid is either placed early or at the last second. There is no psychology or emotion involved there. Agree or disagree?
If you want to stick to the methodology in this study, fine. Although your statement seemed more general. But your point is still moot. As I pointed out. If you bid $100 on day one and your win price is $80 and you bid THE SAME $100 on the final day and your win price is $90. And the only difference is the timing of the bid, then bidding late cost you $10. It IS POSSIBLE, contrary to your assertion, for snipe bidding to cost you money if you are looking at price paid because...
You can NOT ignore the psychology of the other bidders even if there is no psychology to your identical $100 bids. As you mention here...
Now consider the psychology of the other bidders in the auction. If you know anything about poker and game theory, then you know that the less your opponent knows about your intentions, the more advantageous for you. By bidding at the last possible moment, you deprive the opponent(s) of the time to react to your bid. By bidding at the last second, you also eliminate the possibility of shills probing your early bid with incremental raises.
Your "opponent" does not know what your max bid is if you bid on day one. Yes, you eliminate the possibility of shills probing your early bid, assuming that's even common. But you might also consider that you change the whole nature of the auction by eliminating weaker hands early. Overall activity is, according to another study, a factor in where snipers tend to snipe.
You also need to consider the increments. Let's say your "opponent" will bid no higher than $80. If you bid early and get to $80 first, he does not place another bid. On the other hand, if he places his $80 bid first, your $100 snipe forces you to pay $82 or $85 whatever the bid increment is. That is a very simple way that the snipe would cost you MORE even with identical behavior from all bidders.
None of this is an argument against the findings of the study, which are what they are.
Why don't you just admit that you either misspoke or were incorrect? First, you suggested that the statistically insignificant difference was somehow significant. When that didn't work, you suggested that there is no way you ever pay more with a snipe than an early bid, which I've also demonstrated is simply untrue.
It's okay, we still love you even if you're wrong.
I did not misspeak. You keep coming up with off the wall scenarios to try and save face. For example, if I bid my max early, I do not force out weaker hands unless someone else has already entered a bid higher than what they (the weaker hands) wanted to bid. In which case, the weaker hands were never going to be a factor in the auction anyway.
The only point of yours I agree with is that if you bid the same exact amount as someone else, it is advantageous to have done so first. That is probably why experienced snipers tend to add a small random amount to their max bid rather than use a round number.
None of that is off the wall.
Why do i have to save face? It's not even my study you are arguing with? And you just admitted that you could end up paying more with a snipe due to the increments. So, if nothing else you are wrong on that. And, frankly, the statistics speak for themselves.
But, that's okay, I don't ever expect you to admit that you were wrong...especially if it simultaneously means I was right.
The forum may judge. I am done.
You composed the title ("bid sniping does not work"). Yet the data disclosed in the study showed that snipers saved money on average. Then the authors conclude that the savings weren't statistically significant. OK, so don't publish the damn article until you've run enough trials to achieve statistical significance! By the same token, they certainly didn't prove that sniping costs the bidder money. At worst, it's inconclusive.
I did not admit that you'll end up paying more with a snipe. You'll end up winning slightly less often, but it is impossible to construct a scenario where given the same max bids placed by the other bidders you end up paying more with a snipe than had you bid the same amount first. I dare you to try.
NO, the conclusion of the study was quite the opposite: there were NO STATISTICALLY SIGNIFICANT SAVINGS.
As for delaying publication until you have a statistically significant difference ASSUMES that they would eventually find one. Even if they run 1000 trials, there will still be a statistical test of significance. While the confidence interval will be more narrow, it is never zero. And, if there is no statistically significant difference, there is no statistically significant difference.
He's correct here, @CoinJunkie. In statistics, the lack of a statistically significant result is, itself, significant. What it's saying is that although the numbers are different (sniping saves you money) given the parameters of the trial, the difference is within the error band, and the result could have just as easily said sniping costs you money. This is the same reason that medical trials may say a drug is no better than a placebo even when more patients who received the drug get better. If not enough more get better, the results may be due to randomness (the patients who got the drug were slightly less sick than those who got the placebo) rather than because the drug actually worked.
I agree, in the abstract. However, sniping wouldn't be expected to save much money with commonplace, easily valued items (as many have said). And as you noted, there are any number of flaws in the methodology of the study. And most importantly, I have made arguments above for why sniping cannot ever result in a higher final price to the winning bidder than had he/she bid the same amount at the beginning of the auction, but might result in a lower final price. Those arguments have not been refuted, although there has been a lot of babble about psychology, etc.
ACTUALLY THEY HAVE BEEN REFUTED. You just refuse to acknowledge it.
Bit $100 on day 1. Underbidder won't go past $80. You are the first one to $80, you win.
You wait until the last day. Underbidder has already bid $80. You have to go one increment higher.
I can't make it any simpler than that. Even if EVERYONE bids exactly the same - no psychology or anything - you CAN pay more by sniping.
This is incorrect. The winning bid would be $81 either way. If you bid $100 first and then the underbidder won't go past $80 and bids that amount, ebay will automatically bump your bid to $81 if you have previously bid $100 as your high bid. That is how proxy bidding works. In the latter case, the underbidder bids the same $80 and with your snipe, you outbid his $80 and win the item at the next bid increment of $81.
Yes, you are correct with the bidding agent. You'd need a 3rd bidder for you to end up at $80. I make a friendly amendment. LOL.
You need a bidder at $79 and another bidder who refuses to go above $80.
All comments reflect the opinion of the author, even when irrefutably accurate.
@jmlanzaf said:
It's hard to run the test with non-widgets. If you can't pair the "identical" coins, how do you know whether the resulting price difference is due to the timing of the bid or the coin itself?
I think the bigger problem with sniping on "unique" coins is that you have the possibility of two people going nuclear. It must have happened here or there, don't you think?
Agree that it is harder to run the test with non-widgets. Frankly that is obvious. The closer you get to unique, the harder it is to make a valid comparison.
Nuclear bids are part of the game. I have been on both sides of a nuclear bid, but it depends on what your definition of nuclear bid is. If a nuclear bid is defined 50% over market, then I have lost most of my nuclear bids and won a few. But I never, ever bid more than my "walkaway" price. I win some and I lose some, and I don't get upset with a loss but I enjoy a win. However, others may bid way more than their "walkaway price" and either return the item or decide they want to be buried in it for a very long time. I've seen both as a seller and a buyer. I know it is frustrating for sellers sometimes, but that is the game we play.
“In matters of style, swim with the current; in matters of principle, stand like a rock." - Thomas Jefferson
@jmlanzaf said:
It's hard to run the test with non-widgets. If you can't pair the "identical" coins, how do you know whether the resulting price difference is due to the timing of the bid or the coin itself?
I think the bigger problem with sniping on "unique" coins is that you have the possibility of two people going nuclear. It must have happened here or there, don't you think?
Agree that it is harder to run the test with non-widgets. Frankly that is obvious. The closer you get to unique, the harder it is to make a valid comparison.
Nuclear bids are part of the game. I have been on both sides of a nuclear bid, but it depends on what your definition of nuclear bid is. If a nuclear bid is defined 50% over market, then I have lost most of my nuclear bids and won a few. But I never, ever bid more than my "walkaway" price. I win some and I lose some, and I don't get upset with a loss but I enjoy a win. However, others may bid way more than their "walkaway price" and either return the item or decide they want to be buried in it for a very long time. I've seen both as a seller and a buyer. I know it is frustrating for sellers sometimes, but that is the game we play.
I agree. That's why I don't really worry about sniping and other games. I throw done my bid at my number. If I win, I win. If I lose, I lose. If I ever feel like I HAVE to win, I guarantee I'll end up over-paying.
All comments reflect the opinion of the author, even when irrefutably accurate.
@CoinJunkie said:
The conclusion of the article is not that "sniping doesn't work":
In one auction of each pair we submitted an early bid, and in the other we submitted the same bid exactly ten seconds before its close. Our results, from a set of 70 such pairs, indicate no benefit to sniping: we found evidence of 2.54% lower prices for the sniped auctions, but we did not find this benefit to be statistically significant.
In fact, sniping results in slightly lower prices, which is a benefit in spite of what the authors state above. From my perspective, the only downside to sniping is that in a multiway (eBay style) auction, if you are the last person to enter a bid, you can potentially not win the auction even though your bid was the highest. That would be due to your bid not being at least a full increment higher than the current price at the time it was entered.
This is a completely ignorant statement. When the difference is not statistically significant, it means that there is effectively no difference as it is within the margin of error of the measurment.
Use common sense! There is no possible way that sniping can cost you money, but it can save you money. Them's the facts. I don't care about "statistical significance". As I stated, the downside of sniping is that you will very occasionally fail to secure a lot when you in fact submitted the highest bid, so that's the only risk in it.
That's not true. If I submit a nuclear snipe bid to try to win it - which people do - then you can end up with a higher bid than you would normally submit. You actually can't afford to be conservative, because you'll have no chance to counter.
If I submit the same bid as I normally would, no matter what, you are correct. But then I'm less likely to win because the first one to the winning number wins.
Of course, it's not my study, so I don't feel too much passion to defend the conclusion. There are other studies that show slightly different outcomes. None of them seems to show a significant price difference or a significant difference in the likelihood of winning with the sole exception [of studies I found] of a study that showed that if you found unpopular auctions with few bidders you tended to do better. Although, in that case, is the success due to sniping or simple lack of competition?
Nonetheless, I find it interesting. You may not. So be it. But please argue with data not conjecture. Thank you.
People also submit nuclear bids early in auctions. I've seen it happen many times. Just as I've seen many snipe bids of the non-nuclear variety. If you want to turn this into a debate on whether snipe bidders are likely to bid more aggressively than non-snipers, I have no interest. From the paper you cited:
In one auction of each pair we submitted an early bid, and in the other we submitted the same bid exactly ten seconds before its close.
So I believe it was reasonable to assume that that was the scenario under discussion.
And for the record, I am in no way "anti-science".
Your words: ** I don't care about "statistical significance". **
Even if they submit the identical bid early and late, since they don't always win at their maximum bid, it is still possible that the snipe bids ended at a higher number. That, of course, is NOT what the data says. The data, however, also indicates that it didn't end at a lower number.
OK, let me rephrase what I assumed was obvious: I don't care about statistical significance in the context of this argument. I'm not going to rehash everything I've already written. But to summarize: Sniping (vs. making the same bid earlier) can never cost you money, but it can save you money. No statistics are required to prove that argument.
Again, that is only true relative to your max bid. You can't ever pay more than your max bid, sure. But you could make the same case about bidding early. Bidding first can never cost you money, but it could save you money.
>
It is a completely different situation if you are talking about the price paid rather than the max bid.
If bidding $100 on day one leads to an average purchase price of $80 and sniping at $100 leads to an average purchase price of $90 - assuming the type of bid is the ONLY difference - then sniping costs you $10.
Your point is only correct in the limited and moot context of the max bid. No one EVER pays more than their max bid, no matter when placed.
You are assuming that everyone's bid, at any time, is not influenced by what's already happened. You are assuming that everyone is always going to have the same max bid. That may not be true, including the sniper. If you calculate your snipe bid based on the price on day 6 rather than the opening bid on day 1, it may end up higher. There's a whole psychology to this.
There are so many logical fallacies and misdirections in this post, I'm not sure where to begin.
For the purposes of this argument, we are talking about the methodology in the cited study. Namely, the same pre-determined maximum bid is either placed early or at the last second. There is no psychology or emotion involved there. Agree or disagree?
If you want to stick to the methodology in this study, fine. Although your statement seemed more general. But your point is still moot. As I pointed out. If you bid $100 on day one and your win price is $80 and you bid THE SAME $100 on the final day and your win price is $90. And the only difference is the timing of the bid, then bidding late cost you $10. It IS POSSIBLE, contrary to your assertion, for snipe bidding to cost you money if you are looking at price paid because...
You can NOT ignore the psychology of the other bidders even if there is no psychology to your identical $100 bids. As you mention here...
Now consider the psychology of the other bidders in the auction. If you know anything about poker and game theory, then you know that the less your opponent knows about your intentions, the more advantageous for you. By bidding at the last possible moment, you deprive the opponent(s) of the time to react to your bid. By bidding at the last second, you also eliminate the possibility of shills probing your early bid with incremental raises.
Your "opponent" does not know what your max bid is if you bid on day one. Yes, you eliminate the possibility of shills probing your early bid, assuming that's even common. But you might also consider that you change the whole nature of the auction by eliminating weaker hands early. Overall activity is, according to another study, a factor in where snipers tend to snipe.
You also need to consider the increments. Let's say your "opponent" will bid no higher than $80. If you bid early and get to $80 first, he does not place another bid. On the other hand, if he places his $80 bid first, your $100 snipe forces you to pay $82 or $85 whatever the bid increment is. That is a very simple way that the snipe would cost you MORE even with identical behavior from all bidders.
None of this is an argument against the findings of the study, which are what they are.
Why don't you just admit that you either misspoke or were incorrect? First, you suggested that the statistically insignificant difference was somehow significant. When that didn't work, you suggested that there is no way you ever pay more with a snipe than an early bid, which I've also demonstrated is simply untrue.
It's okay, we still love you even if you're wrong.
I did not misspeak. You keep coming up with off the wall scenarios to try and save face. For example, if I bid my max early, I do not force out weaker hands unless someone else has already entered a bid higher than what they (the weaker hands) wanted to bid. In which case, the weaker hands were never going to be a factor in the auction anyway.
The only point of yours I agree with is that if you bid the same exact amount as someone else, it is advantageous to have done so first. That is probably why experienced snipers tend to add a small random amount to their max bid rather than use a round number.
None of that is off the wall.
Why do i have to save face? It's not even my study you are arguing with? And you just admitted that you could end up paying more with a snipe due to the increments. So, if nothing else you are wrong on that. And, frankly, the statistics speak for themselves.
But, that's okay, I don't ever expect you to admit that you were wrong...especially if it simultaneously means I was right.
The forum may judge. I am done.
You composed the title ("bid sniping does not work"). Yet the data disclosed in the study showed that snipers saved money on average. Then the authors conclude that the savings weren't statistically significant. OK, so don't publish the damn article until you've run enough trials to achieve statistical significance! By the same token, they certainly didn't prove that sniping costs the bidder money. At worst, it's inconclusive.
I did not admit that you'll end up paying more with a snipe. You'll end up winning slightly less often, but it is impossible to construct a scenario where given the same max bids placed by the other bidders you end up paying more with a snipe than had you bid the same amount first. I dare you to try.
NO, the conclusion of the study was quite the opposite: there were NO STATISTICALLY SIGNIFICANT SAVINGS.
As for delaying publication until you have a statistically significant difference ASSUMES that they would eventually find one. Even if they run 1000 trials, there will still be a statistical test of significance. While the confidence interval will be more narrow, it is never zero. And, if there is no statistically significant difference, there is no statistically significant difference.
He's correct here, @CoinJunkie. In statistics, the lack of a statistically significant result is, itself, significant. What it's saying is that although the numbers are different (sniping saves you money) given the parameters of the trial, the difference is within the error band, and the result could have just as easily said sniping costs you money. This is the same reason that medical trials may say a drug is no better than a placebo even when more patients who received the drug get better. If not enough more get better, the results may be due to randomness (the patients who got the drug were slightly less sick than those who got the placebo) rather than because the drug actually worked.
I agree, in the abstract. However, sniping wouldn't be expected to save much money with commonplace, easily valued items (as many have said). And as you noted, there are any number of flaws in the methodology of the study. And most importantly, I have made arguments above for why sniping cannot ever result in a higher final price to the winning bidder than had he/she bid the same amount at the beginning of the auction, but might result in a lower final price. Those arguments have not been refuted, although there has been a lot of babble about psychology, etc.
ACTUALLY THEY HAVE BEEN REFUTED. You just refuse to acknowledge it.
Bit $100 on day 1. Underbidder won't go past $80. You are the first one to $80, you win.
You wait until the last day. Underbidder has already bid $80. You have to go one increment higher.
I can't make it any simpler than that. Even if EVERYONE bids exactly the same - no psychology or anything - you CAN pay more by sniping.
This is incorrect. The winning bid would be $81 either way. If you bid $100 first and then the underbidder won't go past $80 and bids that amount, ebay will automatically bump your bid to $81 if you have previously bid $100 as your high bid. That is how proxy bidding works. In the latter case, the underbidder bids the same $80 and with your snipe, you outbid his $80 and win the item at the next bid increment of $81.
Yes, you are correct with the bidding agent. You'd need a 3rd bidder for you to end up at $80. I make a friendly amendment. LOL.
You need a bidder at $79 and another bidder who refuses to go above $80.
Agreed. That is an edge case that would cost you more when sniping. It is unlikely enough and not particularly costly such that it doesn't change my belief that sniping saves the bidder money overall, but I will admit I missed it.
@CoinJunkie said:
The conclusion of the article is not that "sniping doesn't work":
In one auction of each pair we submitted an early bid, and in the other we submitted the same bid exactly ten seconds before its close. Our results, from a set of 70 such pairs, indicate no benefit to sniping: we found evidence of 2.54% lower prices for the sniped auctions, but we did not find this benefit to be statistically significant.
In fact, sniping results in slightly lower prices, which is a benefit in spite of what the authors state above. From my perspective, the only downside to sniping is that in a multiway (eBay style) auction, if you are the last person to enter a bid, you can potentially not win the auction even though your bid was the highest. That would be due to your bid not being at least a full increment higher than the current price at the time it was entered.
This is a completely ignorant statement. When the difference is not statistically significant, it means that there is effectively no difference as it is within the margin of error of the measurment.
Use common sense! There is no possible way that sniping can cost you money, but it can save you money. Them's the facts. I don't care about "statistical significance". As I stated, the downside of sniping is that you will very occasionally fail to secure a lot when you in fact submitted the highest bid, so that's the only risk in it.
That's not true. If I submit a nuclear snipe bid to try to win it - which people do - then you can end up with a higher bid than you would normally submit. You actually can't afford to be conservative, because you'll have no chance to counter.
If I submit the same bid as I normally would, no matter what, you are correct. But then I'm less likely to win because the first one to the winning number wins.
Of course, it's not my study, so I don't feel too much passion to defend the conclusion. There are other studies that show slightly different outcomes. None of them seems to show a significant price difference or a significant difference in the likelihood of winning with the sole exception [of studies I found] of a study that showed that if you found unpopular auctions with few bidders you tended to do better. Although, in that case, is the success due to sniping or simple lack of competition?
Nonetheless, I find it interesting. You may not. So be it. But please argue with data not conjecture. Thank you.
People also submit nuclear bids early in auctions. I've seen it happen many times. Just as I've seen many snipe bids of the non-nuclear variety. If you want to turn this into a debate on whether snipe bidders are likely to bid more aggressively than non-snipers, I have no interest. From the paper you cited:
In one auction of each pair we submitted an early bid, and in the other we submitted the same bid exactly ten seconds before its close.
So I believe it was reasonable to assume that that was the scenario under discussion.
And for the record, I am in no way "anti-science".
Your words: ** I don't care about "statistical significance". **
Even if they submit the identical bid early and late, since they don't always win at their maximum bid, it is still possible that the snipe bids ended at a higher number. That, of course, is NOT what the data says. The data, however, also indicates that it didn't end at a lower number.
OK, let me rephrase what I assumed was obvious: I don't care about statistical significance in the context of this argument. I'm not going to rehash everything I've already written. But to summarize: Sniping (vs. making the same bid earlier) can never cost you money, but it can save you money. No statistics are required to prove that argument.
Again, that is only true relative to your max bid. You can't ever pay more than your max bid, sure. But you could make the same case about bidding early. Bidding first can never cost you money, but it could save you money.
>
It is a completely different situation if you are talking about the price paid rather than the max bid.
If bidding $100 on day one leads to an average purchase price of $80 and sniping at $100 leads to an average purchase price of $90 - assuming the type of bid is the ONLY difference - then sniping costs you $10.
Your point is only correct in the limited and moot context of the max bid. No one EVER pays more than their max bid, no matter when placed.
You are assuming that everyone's bid, at any time, is not influenced by what's already happened. You are assuming that everyone is always going to have the same max bid. That may not be true, including the sniper. If you calculate your snipe bid based on the price on day 6 rather than the opening bid on day 1, it may end up higher. There's a whole psychology to this.
There are so many logical fallacies and misdirections in this post, I'm not sure where to begin.
For the purposes of this argument, we are talking about the methodology in the cited study. Namely, the same pre-determined maximum bid is either placed early or at the last second. There is no psychology or emotion involved there. Agree or disagree?
If you want to stick to the methodology in this study, fine. Although your statement seemed more general. But your point is still moot. As I pointed out. If you bid $100 on day one and your win price is $80 and you bid THE SAME $100 on the final day and your win price is $90. And the only difference is the timing of the bid, then bidding late cost you $10. It IS POSSIBLE, contrary to your assertion, for snipe bidding to cost you money if you are looking at price paid because...
You can NOT ignore the psychology of the other bidders even if there is no psychology to your identical $100 bids. As you mention here...
Now consider the psychology of the other bidders in the auction. If you know anything about poker and game theory, then you know that the less your opponent knows about your intentions, the more advantageous for you. By bidding at the last possible moment, you deprive the opponent(s) of the time to react to your bid. By bidding at the last second, you also eliminate the possibility of shills probing your early bid with incremental raises.
Your "opponent" does not know what your max bid is if you bid on day one. Yes, you eliminate the possibility of shills probing your early bid, assuming that's even common. But you might also consider that you change the whole nature of the auction by eliminating weaker hands early. Overall activity is, according to another study, a factor in where snipers tend to snipe.
You also need to consider the increments. Let's say your "opponent" will bid no higher than $80. If you bid early and get to $80 first, he does not place another bid. On the other hand, if he places his $80 bid first, your $100 snipe forces you to pay $82 or $85 whatever the bid increment is. That is a very simple way that the snipe would cost you MORE even with identical behavior from all bidders.
None of this is an argument against the findings of the study, which are what they are.
Why don't you just admit that you either misspoke or were incorrect? First, you suggested that the statistically insignificant difference was somehow significant. When that didn't work, you suggested that there is no way you ever pay more with a snipe than an early bid, which I've also demonstrated is simply untrue.
It's okay, we still love you even if you're wrong.
I did not misspeak. You keep coming up with off the wall scenarios to try and save face. For example, if I bid my max early, I do not force out weaker hands unless someone else has already entered a bid higher than what they (the weaker hands) wanted to bid. In which case, the weaker hands were never going to be a factor in the auction anyway.
The only point of yours I agree with is that if you bid the same exact amount as someone else, it is advantageous to have done so first. That is probably why experienced snipers tend to add a small random amount to their max bid rather than use a round number.
None of that is off the wall.
Why do i have to save face? It's not even my study you are arguing with? And you just admitted that you could end up paying more with a snipe due to the increments. So, if nothing else you are wrong on that. And, frankly, the statistics speak for themselves.
But, that's okay, I don't ever expect you to admit that you were wrong...especially if it simultaneously means I was right.
The forum may judge. I am done.
You composed the title ("bid sniping does not work"). Yet the data disclosed in the study showed that snipers saved money on average. Then the authors conclude that the savings weren't statistically significant. OK, so don't publish the damn article until you've run enough trials to achieve statistical significance! By the same token, they certainly didn't prove that sniping costs the bidder money. At worst, it's inconclusive.
I did not admit that you'll end up paying more with a snipe. You'll end up winning slightly less often, but it is impossible to construct a scenario where given the same max bids placed by the other bidders you end up paying more with a snipe than had you bid the same amount first. I dare you to try.
NO, the conclusion of the study was quite the opposite: there were NO STATISTICALLY SIGNIFICANT SAVINGS.
As for delaying publication until you have a statistically significant difference ASSUMES that they would eventually find one. Even if they run 1000 trials, there will still be a statistical test of significance. While the confidence interval will be more narrow, it is never zero. And, if there is no statistically significant difference, there is no statistically significant difference.
He's correct here, @CoinJunkie. In statistics, the lack of a statistically significant result is, itself, significant. What it's saying is that although the numbers are different (sniping saves you money) given the parameters of the trial, the difference is within the error band, and the result could have just as easily said sniping costs you money. This is the same reason that medical trials may say a drug is no better than a placebo even when more patients who received the drug get better. If not enough more get better, the results may be due to randomness (the patients who got the drug were slightly less sick than those who got the placebo) rather than because the drug actually worked.
I agree, in the abstract. However, sniping wouldn't be expected to save much money with commonplace, easily valued items (as many have said). And as you noted, there are any number of flaws in the methodology of the study. And most importantly, I have made arguments above for why sniping cannot ever result in a higher final price to the winning bidder than had he/she bid the same amount at the beginning of the auction, but might result in a lower final price. Those arguments have not been refuted, although there has been a lot of babble about psychology, etc.
ACTUALLY THEY HAVE BEEN REFUTED. You just refuse to acknowledge it.
Bit $100 on day 1. Underbidder won't go past $80. You are the first one to $80, you win.
You wait until the last day. Underbidder has already bid $80. You have to go one increment higher.
I can't make it any simpler than that. Even if EVERYONE bids exactly the same - no psychology or anything - you CAN pay more by sniping.
This is incorrect. The winning bid would be $81 either way. If you bid $100 first and then the underbidder won't go past $80 and bids that amount, ebay will automatically bump your bid to $81 if you have previously bid $100 as your high bid. That is how proxy bidding works. In the latter case, the underbidder bids the same $80 and with your snipe, you outbid his $80 and win the item at the next bid increment of $81.
Yes, you are correct with the bidding agent. You'd need a 3rd bidder for you to end up at $80. I make a friendly amendment. LOL.
You need a bidder at $79 and another bidder who refuses to go above $80.
Agreed. That is an edge case that would cost you more when sniping. It is unlikely enough and not particularly costly such that it doesn't change my belief that sniping saves the bidder money overall, but I will admit I missed it.
We agree.
Let's stop there.
MODERATORS, CLOSE THIS THREAD!
All comments reflect the opinion of the author, even when irrefutably accurate.
@CoinJunkie said:
The conclusion of the article is not that "sniping doesn't work":
In one auction of each pair we submitted an early bid, and in the other we submitted the same bid exactly ten seconds before its close. Our results, from a set of 70 such pairs, indicate no benefit to sniping: we found evidence of 2.54% lower prices for the sniped auctions, but we did not find this benefit to be statistically significant.
In fact, sniping results in slightly lower prices, which is a benefit in spite of what the authors state above. From my perspective, the only downside to sniping is that in a multiway (eBay style) auction, if you are the last person to enter a bid, you can potentially not win the auction even though your bid was the highest. That would be due to your bid not being at least a full increment higher than the current price at the time it was entered.
This is a completely ignorant statement. When the difference is not statistically significant, it means that there is effectively no difference as it is within the margin of error of the measurment.
Use common sense! There is no possible way that sniping can cost you money, but it can save you money. Them's the facts. I don't care about "statistical significance". As I stated, the downside of sniping is that you will very occasionally fail to secure a lot when you in fact submitted the highest bid, so that's the only risk in it.
That's not true. If I submit a nuclear snipe bid to try to win it - which people do - then you can end up with a higher bid than you would normally submit. You actually can't afford to be conservative, because you'll have no chance to counter.
If I submit the same bid as I normally would, no matter what, you are correct. But then I'm less likely to win because the first one to the winning number wins.
Of course, it's not my study, so I don't feel too much passion to defend the conclusion. There are other studies that show slightly different outcomes. None of them seems to show a significant price difference or a significant difference in the likelihood of winning with the sole exception [of studies I found] of a study that showed that if you found unpopular auctions with few bidders you tended to do better. Although, in that case, is the success due to sniping or simple lack of competition?
Nonetheless, I find it interesting. You may not. So be it. But please argue with data not conjecture. Thank you.
People also submit nuclear bids early in auctions. I've seen it happen many times. Just as I've seen many snipe bids of the non-nuclear variety. If you want to turn this into a debate on whether snipe bidders are likely to bid more aggressively than non-snipers, I have no interest. From the paper you cited:
In one auction of each pair we submitted an early bid, and in the other we submitted the same bid exactly ten seconds before its close.
So I believe it was reasonable to assume that that was the scenario under discussion.
And for the record, I am in no way "anti-science".
Your words: ** I don't care about "statistical significance". **
Even if they submit the identical bid early and late, since they don't always win at their maximum bid, it is still possible that the snipe bids ended at a higher number. That, of course, is NOT what the data says. The data, however, also indicates that it didn't end at a lower number.
OK, let me rephrase what I assumed was obvious: I don't care about statistical significance in the context of this argument. I'm not going to rehash everything I've already written. But to summarize: Sniping (vs. making the same bid earlier) can never cost you money, but it can save you money. No statistics are required to prove that argument.
Again, that is only true relative to your max bid. You can't ever pay more than your max bid, sure. But you could make the same case about bidding early. Bidding first can never cost you money, but it could save you money.
>
It is a completely different situation if you are talking about the price paid rather than the max bid.
If bidding $100 on day one leads to an average purchase price of $80 and sniping at $100 leads to an average purchase price of $90 - assuming the type of bid is the ONLY difference - then sniping costs you $10.
Your point is only correct in the limited and moot context of the max bid. No one EVER pays more than their max bid, no matter when placed.
You are assuming that everyone's bid, at any time, is not influenced by what's already happened. You are assuming that everyone is always going to have the same max bid. That may not be true, including the sniper. If you calculate your snipe bid based on the price on day 6 rather than the opening bid on day 1, it may end up higher. There's a whole psychology to this.
There are so many logical fallacies and misdirections in this post, I'm not sure where to begin.
For the purposes of this argument, we are talking about the methodology in the cited study. Namely, the same pre-determined maximum bid is either placed early or at the last second. There is no psychology or emotion involved there. Agree or disagree?
If you want to stick to the methodology in this study, fine. Although your statement seemed more general. But your point is still moot. As I pointed out. If you bid $100 on day one and your win price is $80 and you bid THE SAME $100 on the final day and your win price is $90. And the only difference is the timing of the bid, then bidding late cost you $10. It IS POSSIBLE, contrary to your assertion, for snipe bidding to cost you money if you are looking at price paid because...
You can NOT ignore the psychology of the other bidders even if there is no psychology to your identical $100 bids. As you mention here...
Now consider the psychology of the other bidders in the auction. If you know anything about poker and game theory, then you know that the less your opponent knows about your intentions, the more advantageous for you. By bidding at the last possible moment, you deprive the opponent(s) of the time to react to your bid. By bidding at the last second, you also eliminate the possibility of shills probing your early bid with incremental raises.
Your "opponent" does not know what your max bid is if you bid on day one. Yes, you eliminate the possibility of shills probing your early bid, assuming that's even common. But you might also consider that you change the whole nature of the auction by eliminating weaker hands early. Overall activity is, according to another study, a factor in where snipers tend to snipe.
You also need to consider the increments. Let's say your "opponent" will bid no higher than $80. If you bid early and get to $80 first, he does not place another bid. On the other hand, if he places his $80 bid first, your $100 snipe forces you to pay $82 or $85 whatever the bid increment is. That is a very simple way that the snipe would cost you MORE even with identical behavior from all bidders.
None of this is an argument against the findings of the study, which are what they are.
Why don't you just admit that you either misspoke or were incorrect? First, you suggested that the statistically insignificant difference was somehow significant. When that didn't work, you suggested that there is no way you ever pay more with a snipe than an early bid, which I've also demonstrated is simply untrue.
It's okay, we still love you even if you're wrong.
I did not misspeak. You keep coming up with off the wall scenarios to try and save face. For example, if I bid my max early, I do not force out weaker hands unless someone else has already entered a bid higher than what they (the weaker hands) wanted to bid. In which case, the weaker hands were never going to be a factor in the auction anyway.
The only point of yours I agree with is that if you bid the same exact amount as someone else, it is advantageous to have done so first. That is probably why experienced snipers tend to add a small random amount to their max bid rather than use a round number.
None of that is off the wall.
Why do i have to save face? It's not even my study you are arguing with? And you just admitted that you could end up paying more with a snipe due to the increments. So, if nothing else you are wrong on that. And, frankly, the statistics speak for themselves.
But, that's okay, I don't ever expect you to admit that you were wrong...especially if it simultaneously means I was right.
The forum may judge. I am done.
You composed the title ("bid sniping does not work"). Yet the data disclosed in the study showed that snipers saved money on average. Then the authors conclude that the savings weren't statistically significant. OK, so don't publish the damn article until you've run enough trials to achieve statistical significance! By the same token, they certainly didn't prove that sniping costs the bidder money. At worst, it's inconclusive.
I did not admit that you'll end up paying more with a snipe. You'll end up winning slightly less often, but it is impossible to construct a scenario where given the same max bids placed by the other bidders you end up paying more with a snipe than had you bid the same amount first. I dare you to try.
NO, the conclusion of the study was quite the opposite: there were NO STATISTICALLY SIGNIFICANT SAVINGS.
As for delaying publication until you have a statistically significant difference ASSUMES that they would eventually find one. Even if they run 1000 trials, there will still be a statistical test of significance. While the confidence interval will be more narrow, it is never zero. And, if there is no statistically significant difference, there is no statistically significant difference.
He's correct here, @CoinJunkie. In statistics, the lack of a statistically significant result is, itself, significant. What it's saying is that although the numbers are different (sniping saves you money) given the parameters of the trial, the difference is within the error band, and the result could have just as easily said sniping costs you money. This is the same reason that medical trials may say a drug is no better than a placebo even when more patients who received the drug get better. If not enough more get better, the results may be due to randomness (the patients who got the drug were slightly less sick than those who got the placebo) rather than because the drug actually worked.
I agree, in the abstract. However, sniping wouldn't be expected to save much money with commonplace, easily valued items (as many have said). And as you noted, there are any number of flaws in the methodology of the study. And most importantly, I have made arguments above for why sniping cannot ever result in a higher final price to the winning bidder than had he/she bid the same amount at the beginning of the auction, but might result in a lower final price. Those arguments have not been refuted, although there has been a lot of babble about psychology, etc.
ACTUALLY THEY HAVE BEEN REFUTED. You just refuse to acknowledge it.
Bit $100 on day 1. Underbidder won't go past $80. You are the first one to $80, you win.
You wait until the last day. Underbidder has already bid $80. You have to go one increment higher.
I can't make it any simpler than that. Even if EVERYONE bids exactly the same - no psychology or anything - you CAN pay more by sniping.
This is incorrect. The winning bid would be $81 either way. If you bid $100 first and then the underbidder won't go past $80 and bids that amount, ebay will automatically bump your bid to $81 if you have previously bid $100 as your high bid. That is how proxy bidding works. In the latter case, the underbidder bids the same $80 and with your snipe, you outbid his $80 and win the item at the next bid increment of $81.
Yes, you are correct with the bidding agent. You'd need a 3rd bidder for you to end up at $80. I make a friendly amendment. LOL.
You need a bidder at $79 and another bidder who refuses to go above $80.
Agreed. That is an edge case that would cost you more when sniping. It is unlikely enough and not particularly costly such that it doesn't change my belief that sniping saves the bidder money overall, but I will admit I missed it.
We agree.
Let's stop there.
MODERATORS, CLOSE THIS THREAD!
Glad you agree with me that sniping is a good idea overall. Now maybe you should modify the thread title.
Sniping is the only way I’ll ever bid and I’ve bid in 100’s of auctions on EBay. There are only 2 outcomes. I win the item for the amount I’m will to pay or less. Or I lose the item to someone willing to pay more. In live in person auctions when I am purchasing art for my collection I bid up to the amount I’m willing to pay. 2 out comes again. I win or I lose. Getting into a bidding war makes no sense whatsoever. You may win an item and have buyers remorse for way, way over paying.
W.C.Fields "I spent 50% of my money on alcohol, women, and gambling. The other half I wasted.
If the early bid and snipe are the same amount they only matter in relation to the bids of the other bidders. So the question would be how do the bidding strategies effect the other bidder’s strategies.
With the snipe, the only possible impact on another bidder would be to guard against a snipe.
I guess I don’t see how sniping would lead to the other bidders bidding more than if the bid were placed early.
@CoinJunkie said:
The conclusion of the article is not that "sniping doesn't work":
In one auction of each pair we submitted an early bid, and in the other we submitted the same bid exactly ten seconds before its close. Our results, from a set of 70 such pairs, indicate no benefit to sniping: we found evidence of 2.54% lower prices for the sniped auctions, but we did not find this benefit to be statistically significant.
In fact, sniping results in slightly lower prices, which is a benefit in spite of what the authors state above. From my perspective, the only downside to sniping is that in a multiway (eBay style) auction, if you are the last person to enter a bid, you can potentially not win the auction even though your bid was the highest. That would be due to your bid not being at least a full increment higher than the current price at the time it was entered.
This is a completely ignorant statement. When the difference is not statistically significant, it means that there is effectively no difference as it is within the margin of error of the measurment.
Use common sense! There is no possible way that sniping can cost you money, but it can save you money. Them's the facts. I don't care about "statistical significance". As I stated, the downside of sniping is that you will very occasionally fail to secure a lot when you in fact submitted the highest bid, so that's the only risk in it.
That's not true. If I submit a nuclear snipe bid to try to win it - which people do - then you can end up with a higher bid than you would normally submit. You actually can't afford to be conservative, because you'll have no chance to counter.
If I submit the same bid as I normally would, no matter what, you are correct. But then I'm less likely to win because the first one to the winning number wins.
Of course, it's not my study, so I don't feel too much passion to defend the conclusion. There are other studies that show slightly different outcomes. None of them seems to show a significant price difference or a significant difference in the likelihood of winning with the sole exception [of studies I found] of a study that showed that if you found unpopular auctions with few bidders you tended to do better. Although, in that case, is the success due to sniping or simple lack of competition?
Nonetheless, I find it interesting. You may not. So be it. But please argue with data not conjecture. Thank you.
People also submit nuclear bids early in auctions. I've seen it happen many times. Just as I've seen many snipe bids of the non-nuclear variety. If you want to turn this into a debate on whether snipe bidders are likely to bid more aggressively than non-snipers, I have no interest. From the paper you cited:
In one auction of each pair we submitted an early bid, and in the other we submitted the same bid exactly ten seconds before its close.
So I believe it was reasonable to assume that that was the scenario under discussion.
And for the record, I am in no way "anti-science".
Your words: ** I don't care about "statistical significance". **
Even if they submit the identical bid early and late, since they don't always win at their maximum bid, it is still possible that the snipe bids ended at a higher number. That, of course, is NOT what the data says. The data, however, also indicates that it didn't end at a lower number.
OK, let me rephrase what I assumed was obvious: I don't care about statistical significance in the context of this argument. I'm not going to rehash everything I've already written. But to summarize: Sniping (vs. making the same bid earlier) can never cost you money, but it can save you money. No statistics are required to prove that argument.
Again, that is only true relative to your max bid. You can't ever pay more than your max bid, sure. But you could make the same case about bidding early. Bidding first can never cost you money, but it could save you money.
>
It is a completely different situation if you are talking about the price paid rather than the max bid.
If bidding $100 on day one leads to an average purchase price of $80 and sniping at $100 leads to an average purchase price of $90 - assuming the type of bid is the ONLY difference - then sniping costs you $10.
Your point is only correct in the limited and moot context of the max bid. No one EVER pays more than their max bid, no matter when placed.
You are assuming that everyone's bid, at any time, is not influenced by what's already happened. You are assuming that everyone is always going to have the same max bid. That may not be true, including the sniper. If you calculate your snipe bid based on the price on day 6 rather than the opening bid on day 1, it may end up higher. There's a whole psychology to this.
There are so many logical fallacies and misdirections in this post, I'm not sure where to begin.
For the purposes of this argument, we are talking about the methodology in the cited study. Namely, the same pre-determined maximum bid is either placed early or at the last second. There is no psychology or emotion involved there. Agree or disagree?
If you want to stick to the methodology in this study, fine. Although your statement seemed more general. But your point is still moot. As I pointed out. If you bid $100 on day one and your win price is $80 and you bid THE SAME $100 on the final day and your win price is $90. And the only difference is the timing of the bid, then bidding late cost you $10. It IS POSSIBLE, contrary to your assertion, for snipe bidding to cost you money if you are looking at price paid because...
You can NOT ignore the psychology of the other bidders even if there is no psychology to your identical $100 bids. As you mention here...
Now consider the psychology of the other bidders in the auction. If you know anything about poker and game theory, then you know that the less your opponent knows about your intentions, the more advantageous for you. By bidding at the last possible moment, you deprive the opponent(s) of the time to react to your bid. By bidding at the last second, you also eliminate the possibility of shills probing your early bid with incremental raises.
Your "opponent" does not know what your max bid is if you bid on day one. Yes, you eliminate the possibility of shills probing your early bid, assuming that's even common. But you might also consider that you change the whole nature of the auction by eliminating weaker hands early. Overall activity is, according to another study, a factor in where snipers tend to snipe.
You also need to consider the increments. Let's say your "opponent" will bid no higher than $80. If you bid early and get to $80 first, he does not place another bid. On the other hand, if he places his $80 bid first, your $100 snipe forces you to pay $82 or $85 whatever the bid increment is. That is a very simple way that the snipe would cost you MORE even with identical behavior from all bidders.
None of this is an argument against the findings of the study, which are what they are.
Why don't you just admit that you either misspoke or were incorrect? First, you suggested that the statistically insignificant difference was somehow significant. When that didn't work, you suggested that there is no way you ever pay more with a snipe than an early bid, which I've also demonstrated is simply untrue.
It's okay, we still love you even if you're wrong.
I did not misspeak. You keep coming up with off the wall scenarios to try and save face. For example, if I bid my max early, I do not force out weaker hands unless someone else has already entered a bid higher than what they (the weaker hands) wanted to bid. In which case, the weaker hands were never going to be a factor in the auction anyway.
The only point of yours I agree with is that if you bid the same exact amount as someone else, it is advantageous to have done so first. That is probably why experienced snipers tend to add a small random amount to their max bid rather than use a round number.
None of that is off the wall.
Why do i have to save face? It's not even my study you are arguing with? And you just admitted that you could end up paying more with a snipe due to the increments. So, if nothing else you are wrong on that. And, frankly, the statistics speak for themselves.
But, that's okay, I don't ever expect you to admit that you were wrong...especially if it simultaneously means I was right.
The forum may judge. I am done.
You composed the title ("bid sniping does not work"). Yet the data disclosed in the study showed that snipers saved money on average. Then the authors conclude that the savings weren't statistically significant. OK, so don't publish the damn article until you've run enough trials to achieve statistical significance! By the same token, they certainly didn't prove that sniping costs the bidder money. At worst, it's inconclusive.
I did not admit that you'll end up paying more with a snipe. You'll end up winning slightly less often, but it is impossible to construct a scenario where given the same max bids placed by the other bidders you end up paying more with a snipe than had you bid the same amount first. I dare you to try.
NO, the conclusion of the study was quite the opposite: there were NO STATISTICALLY SIGNIFICANT SAVINGS.
As for delaying publication until you have a statistically significant difference ASSUMES that they would eventually find one. Even if they run 1000 trials, there will still be a statistical test of significance. While the confidence interval will be more narrow, it is never zero. And, if there is no statistically significant difference, there is no statistically significant difference.
He's correct here, @CoinJunkie. In statistics, the lack of a statistically significant result is, itself, significant. What it's saying is that although the numbers are different (sniping saves you money) given the parameters of the trial, the difference is within the error band, and the result could have just as easily said sniping costs you money. This is the same reason that medical trials may say a drug is no better than a placebo even when more patients who received the drug get better. If not enough more get better, the results may be due to randomness (the patients who got the drug were slightly less sick than those who got the placebo) rather than because the drug actually worked.
I agree, in the abstract. However, sniping wouldn't be expected to save much money with commonplace, easily valued items (as many have said). And as you noted, there are any number of flaws in the methodology of the study. And most importantly, I have made arguments above for why sniping cannot ever result in a higher final price to the winning bidder than had he/she bid the same amount at the beginning of the auction, but might result in a lower final price. Those arguments have not been refuted, although there has been a lot of babble about psychology, etc.
ACTUALLY THEY HAVE BEEN REFUTED. You just refuse to acknowledge it.
Bit $100 on day 1. Underbidder won't go past $80. You are the first one to $80, you win.
You wait until the last day. Underbidder has already bid $80. You have to go one increment higher.
I can't make it any simpler than that. Even if EVERYONE bids exactly the same - no psychology or anything - you CAN pay more by sniping.
READ the conditions of the thought experiment!!! You make the same max bid either at the very beginning or at the very end of the auction. In the example you present, when you snipe, you LOSE the auction, you do not "have to go one increment higher". I've already pointed out multiple times that the downside to sniping is that you will occasionally lose when your max bid is equal to or higher than any other bidder's.
The logical conclusion is that if you prioritize winning the lot over getting the best possible price, sniping may not (although that is arguable) be a good choice. If getting the best possible price on any lot won is the priority, then sniping is DEMONSTRABLY (without running ANY simulations or empirical studies) your best strategy. The savings probably won't make you rich, but they are savings nonetheless.
Over and out!
That is not true. Dear God, read slower.
You bid $100 in both cases. You WIN In both cases. It costs you $80 when you bid day 1. It costs you $82 when you sniped.
CONTRARY to your claim that sniping can never cost you more.
Q.E.D.
That’s making the assumption that the other bidder wouldn’t raise the bid if you bid earlier.
@CoinJunkie said:
The conclusion of the article is not that "sniping doesn't work":
In one auction of each pair we submitted an early bid, and in the other we submitted the same bid exactly ten seconds before its close. Our results, from a set of 70 such pairs, indicate no benefit to sniping: we found evidence of 2.54% lower prices for the sniped auctions, but we did not find this benefit to be statistically significant.
In fact, sniping results in slightly lower prices, which is a benefit in spite of what the authors state above. From my perspective, the only downside to sniping is that in a multiway (eBay style) auction, if you are the last person to enter a bid, you can potentially not win the auction even though your bid was the highest. That would be due to your bid not being at least a full increment higher than the current price at the time it was entered.
This is a completely ignorant statement. When the difference is not statistically significant, it means that there is effectively no difference as it is within the margin of error of the measurment.
Use common sense! There is no possible way that sniping can cost you money, but it can save you money. Them's the facts. I don't care about "statistical significance". As I stated, the downside of sniping is that you will very occasionally fail to secure a lot when you in fact submitted the highest bid, so that's the only risk in it.
That's not true. If I submit a nuclear snipe bid to try to win it - which people do - then you can end up with a higher bid than you would normally submit. You actually can't afford to be conservative, because you'll have no chance to counter.
If I submit the same bid as I normally would, no matter what, you are correct. But then I'm less likely to win because the first one to the winning number wins.
Of course, it's not my study, so I don't feel too much passion to defend the conclusion. There are other studies that show slightly different outcomes. None of them seems to show a significant price difference or a significant difference in the likelihood of winning with the sole exception [of studies I found] of a study that showed that if you found unpopular auctions with few bidders you tended to do better. Although, in that case, is the success due to sniping or simple lack of competition?
Nonetheless, I find it interesting. You may not. So be it. But please argue with data not conjecture. Thank you.
People also submit nuclear bids early in auctions. I've seen it happen many times. Just as I've seen many snipe bids of the non-nuclear variety. If you want to turn this into a debate on whether snipe bidders are likely to bid more aggressively than non-snipers, I have no interest. From the paper you cited:
In one auction of each pair we submitted an early bid, and in the other we submitted the same bid exactly ten seconds before its close.
So I believe it was reasonable to assume that that was the scenario under discussion.
And for the record, I am in no way "anti-science".
Your words: ** I don't care about "statistical significance". **
Even if they submit the identical bid early and late, since they don't always win at their maximum bid, it is still possible that the snipe bids ended at a higher number. That, of course, is NOT what the data says. The data, however, also indicates that it didn't end at a lower number.
OK, let me rephrase what I assumed was obvious: I don't care about statistical significance in the context of this argument. I'm not going to rehash everything I've already written. But to summarize: Sniping (vs. making the same bid earlier) can never cost you money, but it can save you money. No statistics are required to prove that argument.
Again, that is only true relative to your max bid. You can't ever pay more than your max bid, sure. But you could make the same case about bidding early. Bidding first can never cost you money, but it could save you money.
>
It is a completely different situation if you are talking about the price paid rather than the max bid.
If bidding $100 on day one leads to an average purchase price of $80 and sniping at $100 leads to an average purchase price of $90 - assuming the type of bid is the ONLY difference - then sniping costs you $10.
Your point is only correct in the limited and moot context of the max bid. No one EVER pays more than their max bid, no matter when placed.
You are assuming that everyone's bid, at any time, is not influenced by what's already happened. You are assuming that everyone is always going to have the same max bid. That may not be true, including the sniper. If you calculate your snipe bid based on the price on day 6 rather than the opening bid on day 1, it may end up higher. There's a whole psychology to this.
There are so many logical fallacies and misdirections in this post, I'm not sure where to begin.
For the purposes of this argument, we are talking about the methodology in the cited study. Namely, the same pre-determined maximum bid is either placed early or at the last second. There is no psychology or emotion involved there. Agree or disagree?
If you want to stick to the methodology in this study, fine. Although your statement seemed more general. But your point is still moot. As I pointed out. If you bid $100 on day one and your win price is $80 and you bid THE SAME $100 on the final day and your win price is $90. And the only difference is the timing of the bid, then bidding late cost you $10. It IS POSSIBLE, contrary to your assertion, for snipe bidding to cost you money if you are looking at price paid because...
You can NOT ignore the psychology of the other bidders even if there is no psychology to your identical $100 bids. As you mention here...
Now consider the psychology of the other bidders in the auction. If you know anything about poker and game theory, then you know that the less your opponent knows about your intentions, the more advantageous for you. By bidding at the last possible moment, you deprive the opponent(s) of the time to react to your bid. By bidding at the last second, you also eliminate the possibility of shills probing your early bid with incremental raises.
Your "opponent" does not know what your max bid is if you bid on day one. Yes, you eliminate the possibility of shills probing your early bid, assuming that's even common. But you might also consider that you change the whole nature of the auction by eliminating weaker hands early. Overall activity is, according to another study, a factor in where snipers tend to snipe.
You also need to consider the increments. Let's say your "opponent" will bid no higher than $80. If you bid early and get to $80 first, he does not place another bid. On the other hand, if he places his $80 bid first, your $100 snipe forces you to pay $82 or $85 whatever the bid increment is. That is a very simple way that the snipe would cost you MORE even with identical behavior from all bidders.
None of this is an argument against the findings of the study, which are what they are.
Why don't you just admit that you either misspoke or were incorrect? First, you suggested that the statistically insignificant difference was somehow significant. When that didn't work, you suggested that there is no way you ever pay more with a snipe than an early bid, which I've also demonstrated is simply untrue.
It's okay, we still love you even if you're wrong.
I did not misspeak. You keep coming up with off the wall scenarios to try and save face. For example, if I bid my max early, I do not force out weaker hands unless someone else has already entered a bid higher than what they (the weaker hands) wanted to bid. In which case, the weaker hands were never going to be a factor in the auction anyway.
The only point of yours I agree with is that if you bid the same exact amount as someone else, it is advantageous to have done so first. That is probably why experienced snipers tend to add a small random amount to their max bid rather than use a round number.
None of that is off the wall.
Why do i have to save face? It's not even my study you are arguing with? And you just admitted that you could end up paying more with a snipe due to the increments. So, if nothing else you are wrong on that. And, frankly, the statistics speak for themselves.
But, that's okay, I don't ever expect you to admit that you were wrong...especially if it simultaneously means I was right.
The forum may judge. I am done.
You composed the title ("bid sniping does not work"). Yet the data disclosed in the study showed that snipers saved money on average. Then the authors conclude that the savings weren't statistically significant. OK, so don't publish the damn article until you've run enough trials to achieve statistical significance! By the same token, they certainly didn't prove that sniping costs the bidder money. At worst, it's inconclusive.
I did not admit that you'll end up paying more with a snipe. You'll end up winning slightly less often, but it is impossible to construct a scenario where given the same max bids placed by the other bidders you end up paying more with a snipe than had you bid the same amount first. I dare you to try.
NO, the conclusion of the study was quite the opposite: there were NO STATISTICALLY SIGNIFICANT SAVINGS.
As for delaying publication until you have a statistically significant difference ASSUMES that they would eventually find one. Even if they run 1000 trials, there will still be a statistical test of significance. While the confidence interval will be more narrow, it is never zero. And, if there is no statistically significant difference, there is no statistically significant difference.
He's correct here, @CoinJunkie. In statistics, the lack of a statistically significant result is, itself, significant. What it's saying is that although the numbers are different (sniping saves you money) given the parameters of the trial, the difference is within the error band, and the result could have just as easily said sniping costs you money. This is the same reason that medical trials may say a drug is no better than a placebo even when more patients who received the drug get better. If not enough more get better, the results may be due to randomness (the patients who got the drug were slightly less sick than those who got the placebo) rather than because the drug actually worked.
I agree, in the abstract. However, sniping wouldn't be expected to save much money with commonplace, easily valued items (as many have said). And as you noted, there are any number of flaws in the methodology of the study. And most importantly, I have made arguments above for why sniping cannot ever result in a higher final price to the winning bidder than had he/she bid the same amount at the beginning of the auction, but might result in a lower final price. Those arguments have not been refuted, although there has been a lot of babble about psychology, etc.
ACTUALLY THEY HAVE BEEN REFUTED. You just refuse to acknowledge it.
Bit $100 on day 1. Underbidder won't go past $80. You are the first one to $80, you win.
You wait until the last day. Underbidder has already bid $80. You have to go one increment higher.
I can't make it any simpler than that. Even if EVERYONE bids exactly the same - no psychology or anything - you CAN pay more by sniping.
READ the conditions of the thought experiment!!! You make the same max bid either at the very beginning or at the very end of the auction. In the example you present, when you snipe, you LOSE the auction, you do not "have to go one increment higher". I've already pointed out multiple times that the downside to sniping is that you will occasionally lose when your max bid is equal to or higher than any other bidder's.
The logical conclusion is that if you prioritize winning the lot over getting the best possible price, sniping may not (although that is arguable) be a good choice. If getting the best possible price on any lot won is the priority, then sniping is DEMONSTRABLY (without running ANY simulations or empirical studies) your best strategy. The savings probably won't make you rich, but they are savings nonetheless.
Over and out!
That is not true. Dear God, read slower.
You bid $100 in both cases. You WIN In both cases. It costs you $80 when you bid day 1. It costs you $82 when you sniped.
CONTRARY to your claim that sniping can never cost you more.
Q.E.D.
That’s making the assumption that the other bidder wouldn’t raise the bid if you bid earlier.
This study has lots of issues - and there are a number of 'peer-reviewed' journals out there that publish poorly conducted 'science'.
Further, and they may have stated this but I did not notice because I gave up reading it, the conclusions only apply to the items they studied and in the time period they studied them. So extending their results to snipe/no snipe more broadly doesn't hold.
Etc.,etc....
@ricko said:
They were looking for a cost benefit....When I snipe(d), I was looking for an acquisition benefit...which means, I got the item and the others did not. When evaluated from that position, sniping is good. Cheers, RickO
I agree... I never thought of sniping as a way of saving money.
I only "snipe" on items I really want, and usually they are items that only "in the know" people can determine the true value. You don't want to disclose that kind of stuff with 20 seconds to go to clueless bidders who only want to beat the last bid placed. These items are never generic and not comparable to previous sales.
If you're "sniping" with 10 seconds to go....that's way too early. It's not even a snipe imo. I can react manually to such a bid and still take it out manually with a couple seconds to go. A "manual" snipe should probably be entered no sooner than 4-7 seconds. For those with snipe programs....you have the upper hand.
Go ahead and put up your nuclear bid early and see how often it will get "uncovered" either by shills, 0 FB bidders living in their Mom's basement, or other "nuclear" bidders who now have the edge over you since you just played your hand.
Never snipe in round number bid increments. If $100 then go $100.57. Or if at say $200 go $207.57 or even $211.11. But do include shipping and taxes in your total cost "before" bidding. Some snipers will just snipe on the "bid" number while other smarter ones will snipe on the "total costs" as well....covering both bases.
I do enough sniping for particular items that I can recognize some other higher volume, high FB dealers who routinely snipe for immediate resale. I can tell them by the FB numbers attached to their "hidden" IDs. In that respect it's almost always nice to know you out-sniped a top national player....rather than just a lower feed back newbie who get "too excited"
I get out-sniped 50-75% of the time....almost always from other experts or dealers who know more than I do or who have deeper pockets or better avenues to dispose of their "kills" and/or "losers." It's all part of the equation. I'm happy getting something from time to time.
Bidding Early and winning....means you paid the most money of anyone in the country when all interested bidders had plenty of time to think about it. Sometimes that works and you get a fair deal, about the same as if you had sniped. But most of the time someone will take out your bid in the last minute....if only just to "take out your bid" and show you up....even if they have no intention of paying.
Saw an Ebay item pop up with starting $99 bid with a $250 BIN. It was a scarce item and I wanted it. Most wouldn't have known what it was. I offered the seller $135 for it ($155 shipped and taxed)....sort of a "fair" low ball imo. I felt it was worth $175+.
Seller passed on my offer and wanted to see the item "run." We both agreed it could fetch considerably more than my offer. I check on the item with an hour to go and no bids....still at $99. Very strange. I couldn't explain why not a single bid for a quality, problem free item was not yet placed. Well, I got caught up in surfing that hour and didn't get back in time to put in a "nuclear bid." The item FAILED to sell. No bids. I was flabbergasted and upset that I missed out. It did end on Sunday midnight on the west coast....not a great time for an auction to end.
I got back to the seller and said I'd be willing to go a little over the starting bid since no one bid. We both must have missed something. They offered it to me around my original offer....at $130. I passed.....and said let's just let it run again and both of us take our chances. This time I was ready and placed my max bid of $135 with 5 seconds to go. Got it....for $99....or $115 all in shipped/taxed. No one but me bid on it. Once I got it in hand, I confirmed it was all there and worth in the $150-$175 range. For whatever reason, this one didn't catch anyone else's eye....there were dozen "watchers" on it though. Seller had reasonable FB and at 100%, with low shipping costs. It didn't help that the seller's description was a bit "jumbled" and could have been better.
@btcollects said:
Couple things... increment rules vary at each auction venue, at ebay, a $500.01 bid beats a $500.00 bid, doesn't have to be full increment over. Most other auctions have to be full increment win.
If you try to enter that $500.01 bid in the last few seconds where there is already one in play at $500.00.....the Ebay system will not allow it to be entered. I've had numerous snipe bids rejected in the last couple of seconds because I wasn't a full bid increment higher than a max bid already entered and in play....entered a second before mine. Now if you bid $500.01 well before the bidding has escalated to the $500.00 level (say at $450), then the system may well allow a bid of $500.01 as the $500.00 bid is not in view yet. But I'm not sure of that. To be fair, that person holding the $500 bid should be able to also hold the entire bid increment just above it (ie. $500.00-$509.99). To allow it to be chiseled for $500.01 is not very fair.
@btcollects said:
Couple things... increment rules vary at each auction venue, at ebay, a $500.01 bid beats a $500.00 bid, doesn't have to be full increment over. Most other auctions have to be full increment win.
If you try to enter that $500.01 bid in the last few seconds where there is already one in play at $500.00.....the Ebay system will not allow it to be entered. I've had numerous snipe bids rejected in the last couple of seconds because I wasn't a full bid increment higher than a max bid already entered and in play. Now if you bid $500.01 well before the bidding has escalated to the $500.00 level (say at $450), then the system may well allow a bid of $500.01 as the $500.00 bid is not in view yet. But I'm not sure of that. To be fair, that person holding the $500 bid should be able to also hold the entire bid increment just above it (ie. $500.00-$509.99). To allow it to be chiseled for $500.01 is not very fair.
Any arbitrary amount will be accepted by eBay as long as it's at least a full bid increment higher than the current bid. I believe Heritage is the same in terms of pre-auction Internet bids. Not sure I agree regarding fairness. Max bids are just proxies until they are actualized. If you want to "hold the entire bid increment", increase your max accordingly.
@btcollects said:
Couple things... increment rules vary at each auction venue, at ebay, a $500.01 bid beats a $500.00 bid, doesn't have to be full increment over. Most other auctions have to be full increment win.
If you try to enter that $500.01 bid in the last few seconds where there is already one in play at $500.00.....the Ebay system will not allow it to be entered. I've had numerous snipe bids rejected in the last couple of seconds because I wasn't a full bid increment higher than a max bid already entered and in play. Now if you bid $500.01 well before the bidding has escalated to the $500.00 level (say at $450), then the system may well allow a bid of $500.01 as the $500.00 bid is not in view yet. But I'm not sure of that. To be fair, that person holding the $500 bid should be able to also hold the entire bid increment just above it (ie. $500.00-$509.99). To allow it to be chiseled for $500.01 is not very fair.
Interesting, you learn something new every day. This is a nuanced difference, isn't it? A $501.01 bid can beat a $500.00 bid, but the $500.01 bid has to be accepted. So, when I've won coins for exactly my off-increment max bid, a few cents more than the underbidder in the last few seconds, it's because my manual "snipe" actually was accepted as an over-increment bid, and then the auto bid or true snipe hits and raises my bid to my max. I don't think it's like this with every auction venue.
I'm not sure I understand what you're saying here, but if the current bid is $500, a bid of $501.01 will be rejected just the same as if you'd bid $500.01. If you've successfully entered either $500.01 or $501.01 as a max bid and someone bids $500 at the last second, you'll win in either case for your max.
@CoinJunkie said:
The conclusion of the article is not that "sniping doesn't work":
In one auction of each pair we submitted an early bid, and in the other we submitted the same bid exactly ten seconds before its close. Our results, from a set of 70 such pairs, indicate no benefit to sniping: we found evidence of 2.54% lower prices for the sniped auctions, but we did not find this benefit to be statistically significant.
In fact, sniping results in slightly lower prices, which is a benefit in spite of what the authors state above. From my perspective, the only downside to sniping is that in a multiway (eBay style) auction, if you are the last person to enter a bid, you can potentially not win the auction even though your bid was the highest. That would be due to your bid not being at least a full increment higher than the current price at the time it was entered.
This is a completely ignorant statement. When the difference is not statistically significant, it means that there is effectively no difference as it is within the margin of error of the measurment.
It is interesting that you start your statement by accusing someone else of making a 'completely ignorant statement" then you go on to misstate statistics. If a study concludes that something is not statistically significant, it does not mean that there is effectively no difference. It means that the study was not able to conclude that their was a benefit, because of the limits of sampling. There is a big difference between that and saying their is no difference.
As a matter of fact, the fact that they showed a benefit makes it more likely that there is a benefit to it than not, they just cannot be sure. Think of a political poll, even if you are in the margin of error, the candidate ahead in the poll is more likely to be ahead than the person that was behind in the poll.
@btcollects said:
Couple things... increment rules vary at each auction venue, at ebay, a $500.01 bid beats a $500.00 bid, doesn't have to be full increment over. Most other auctions have to be full increment win.
If you try to enter that $500.01 bid in the last few seconds where there is already one in play at $500.00.....the Ebay system will not allow it to be entered. I've had numerous snipe bids rejected in the last couple of seconds because I wasn't a full bid increment higher than a max bid already entered and in play. Now if you bid $500.01 well before the bidding has escalated to the $500.00 level (say at $450), then the system may well allow a bid of $500.01 as the $500.00 bid is not in view yet. But I'm not sure of that. To be fair, that person holding the $500 bid should be able to also hold the entire bid increment just above it (ie. $500.00-$509.99). To allow it to be chiseled for $500.01 is not very fair.
Interesting, you learn something new every day. This is a nuanced difference, isn't it? A $501.01 bid can beat a $500.00 bid, but the $500.01 bid has to be accepted. So, when I've won coins for exactly my off-increment max bid, a few cents more than the underbidder in the last few seconds, it's because my manual "snipe" actually was accepted as an over-increment bid, and then the auto bid or true snipe hits and raises my bid to my max. I don't think it's like this with every auction venue.
I'm not sure I understand what you're saying here, but if the current bid is $500, a bid of $501.01 will be rejected just the same as if you'd bid $500.01. If you've successfully entered either $500.01 or $501.01 as a max bid and someone bids $500 at the last second, you'll win in either case for your max.
The $501.01 snipe bid only goes off as a snipe if the prevailing high bid is at least one increment below the snipe amount (in this case no higher than $490.01), in which case the snipe of $500.01 will win the auction vs the $500.00 bidder. This is why it's always best not to enter bids at round numbers.
Collecting 1970s Topps baseball wax, rack and cello packs, as well as PCGS graded Half Cents, Large Cents, Two Cent pieces and Three Cent Silver pieces.
@btcollects said:
Couple things... increment rules vary at each auction venue, at ebay, a $500.01 bid beats a $500.00 bid, doesn't have to be full increment over. Most other auctions have to be full increment win.
If you try to enter that $500.01 bid in the last few seconds where there is already one in play at $500.00.....the Ebay system will not allow it to be entered. I've had numerous snipe bids rejected in the last couple of seconds because I wasn't a full bid increment higher than a max bid already entered and in play. Now if you bid $500.01 well before the bidding has escalated to the $500.00 level (say at $450), then the system may well allow a bid of $500.01 as the $500.00 bid is not in view yet. But I'm not sure of that. To be fair, that person holding the $500 bid should be able to also hold the entire bid increment just above it (ie. $500.00-$509.99). To allow it to be chiseled for $500.01 is not very fair.
Interesting, you learn something new every day. This is a nuanced difference, isn't it? A $501.01 bid can beat a $500.00 bid, but the $500.01 bid has to be accepted. So, when I've won coins for exactly my off-increment max bid, a few cents more than the underbidder in the last few seconds, it's because my manual "snipe" actually was accepted as an over-increment bid, and then the auto bid or true snipe hits and raises my bid to my max. I don't think it's like this with every auction venue.
I'm not sure I understand what you're saying here, but if the current bid is $500, a bid of $501.01 will be rejected just the same as if you'd bid $500.01. If you've successfully entered either $500.01 or $501.01 as a max bid and someone bids $500 at the last second, you'll win in either case for your max.
The $501.01 snipe bid only goes off as a snipe if the prevailing high bid is at least one increment below the snipe amount (in this case no higher than $490.01), in which case the snipe of $500.01 will win the auction vs the $500.00 bidder. This is why it's always best not to enter bids at round numbers.
I generally agree that it's best not to enter (snipe) bids at round numbers, but neither $500.01 nor $501.01 is a round number. If the current bid is at $490.01 and that bidder's max is $500, and two last second snipers bid $500.01 and $501.01, respectively, whichever sniper bids first is the winner.
@CoinJunkie said:
The conclusion of the article is not that "sniping doesn't work":
In one auction of each pair we submitted an early bid, and in the other we submitted the same bid exactly ten seconds before its close. Our results, from a set of 70 such pairs, indicate no benefit to sniping: we found evidence of 2.54% lower prices for the sniped auctions, but we did not find this benefit to be statistically significant.
In fact, sniping results in slightly lower prices, which is a benefit in spite of what the authors state above. From my perspective, the only downside to sniping is that in a multiway (eBay style) auction, if you are the last person to enter a bid, you can potentially not win the auction even though your bid was the highest. That would be due to your bid not being at least a full increment higher than the current price at the time it was entered.
This is a completely ignorant statement. When the difference is not statistically significant, it means that there is effectively no difference as it is within the margin of error of the measurment.
It is interesting that you start your statement by accusing someone else of making a 'completely ignorant statement" then you go on to misstate statistics. If a study concludes that something is not statistically significant, it does not mean that there is effectively no difference. It means that the study was not able to conclude that their was a benefit, because of the limits of sampling. There is a big difference between that and saying their is no difference.
As a matter of fact, the fact that they showed a benefit makes it more likely that there is a benefit to it than not, they just cannot be sure. Think of a political poll, even if you are in the margin of error, the candidate ahead in the poll is more likely to be ahead than the person that was behind in the poll.
I said "effectively". A poll that shows a1% difference with a 5% margin of error is not showing a real difference. It is nearly just as likely that the 1% actually goes the other way.
You can NOT ignore the statistical significance of the result in drawing conclusions. If you form the question as a null hypothesis, the result does NOT allow you to conclude there is a benefit. It also, to be fair, does not allow you to conclude there is no benefit.
All comments reflect the opinion of the author, even when irrefutably accurate.
@btcollects said:
Couple things... increment rules vary at each auction venue, at ebay, a $500.01 bid beats a $500.00 bid, doesn't have to be full increment over. Most other auctions have to be full increment win.
If you try to enter that $500.01 bid in the last few seconds where there is already one in play at $500.00.....the Ebay system will not allow it to be entered. I've had numerous snipe bids rejected in the last couple of seconds because I wasn't a full bid increment higher than a max bid already entered and in play. Now if you bid $500.01 well before the bidding has escalated to the $500.00 level (say at $450), then the system may well allow a bid of $500.01 as the $500.00 bid is not in view yet. But I'm not sure of that. To be fair, that person holding the $500 bid should be able to also hold the entire bid increment just above it (ie. $500.00-$509.99). To allow it to be chiseled for $500.01 is not very fair.
Interesting, you learn something new every day. This is a nuanced difference, isn't it? A $501.01 bid can beat a $500.00 bid, but the $500.01 bid has to be accepted. So, when I've won coins for exactly my off-increment max bid, a few cents more than the underbidder in the last few seconds, it's because my manual "snipe" actually was accepted as an over-increment bid, and then the auto bid or true snipe hits and raises my bid to my max. I don't think it's like this with every auction venue.
I'm not sure I understand what you're saying here, but if the current bid is $500, a bid of $501.01 will be rejected just the same as if you'd bid $500.01. If you've successfully entered either $500.01 or $501.01 as a max bid and someone bids $500 at the last second, you'll win in either case for your max.
The $501.01 snipe bid only goes off as a snipe if the prevailing high bid is at least one increment below the snipe amount (in this case no higher than $490.01), in which case the snipe of $500.01 will win the auction vs the $500.00 bidder. This is why it's always best not to enter bids at round numbers.
I generally agree that it's best not to enter (snipe) bids at round numbers, but neither $500.01 nor $501.01 is a round number. If the current bid is at $490.01 and that bidder's max is $500, and two last second snipers bid $500.01 and $501.01, respectively, whichever sniper bids first is the winner.
Agreed. I was talking about a bid of $500.00 even.
Collecting 1970s Topps baseball wax, rack and cello packs, as well as PCGS graded Half Cents, Large Cents, Two Cent pieces and Three Cent Silver pieces.
@CoinJunkie said:
The conclusion of the article is not that "sniping doesn't work":
In one auction of each pair we submitted an early bid, and in the other we submitted the same bid exactly ten seconds before its close. Our results, from a set of 70 such pairs, indicate no benefit to sniping: we found evidence of 2.54% lower prices for the sniped auctions, but we did not find this benefit to be statistically significant.
In fact, sniping results in slightly lower prices, which is a benefit in spite of what the authors state above. From my perspective, the only downside to sniping is that in a multiway (eBay style) auction, if you are the last person to enter a bid, you can potentially not win the auction even though your bid was the highest. That would be due to your bid not being at least a full increment higher than the current price at the time it was entered.
This is a completely ignorant statement. When the difference is not statistically significant, it means that there is effectively no difference as it is within the margin of error of the measurment.
It is interesting that you start your statement by accusing someone else of making a 'completely ignorant statement" then you go on to misstate statistics. If a study concludes that something is not statistically significant, it does not mean that there is effectively no difference. It means that the study was not able to conclude that their was a benefit, because of the limits of sampling. There is a big difference between that and saying their is no difference.
As a matter of fact, the fact that they showed a benefit makes it more likely that there is a benefit to it than not, they just cannot be sure. Think of a political poll, even if you are in the margin of error, the candidate ahead in the poll is more likely to be ahead than the person that was behind in the poll.
I said "effectively". A poll that shows a1% difference with a 5% margin of error is not showing a real difference. It is nearly just as likely that the 1% actually goes the other way.
You can NOT ignore the statistical significance of the result in drawing conclusions. If you form the question as a null hypothesis, the result does NOT allow you to conclude there is a benefit. It also, to be fair, does not allow you to conclude there is no benefit.
And yet your thread title says "showing that bid sniping does not work".
As has already been discussed ad nauseum here, bid sniping on nominally identical items that sell regularly and have a well-established value would not be expected to provide significant benefit. In that sense, the study confirms what is intuitively obvious.
@CoinJunkie said:
The conclusion of the article is not that "sniping doesn't work":
In one auction of each pair we submitted an early bid, and in the other we submitted the same bid exactly ten seconds before its close. Our results, from a set of 70 such pairs, indicate no benefit to sniping: we found evidence of 2.54% lower prices for the sniped auctions, but we did not find this benefit to be statistically significant.
In fact, sniping results in slightly lower prices, which is a benefit in spite of what the authors state above. From my perspective, the only downside to sniping is that in a multiway (eBay style) auction, if you are the last person to enter a bid, you can potentially not win the auction even though your bid was the highest. That would be due to your bid not being at least a full increment higher than the current price at the time it was entered.
This is a completely ignorant statement. When the difference is not statistically significant, it means that there is effectively no difference as it is within the margin of error of the measurment.
It is interesting that you start your statement by accusing someone else of making a 'completely ignorant statement" then you go on to misstate statistics. If a study concludes that something is not statistically significant, it does not mean that there is effectively no difference. It means that the study was not able to conclude that their was a benefit, because of the limits of sampling. There is a big difference between that and saying their is no difference.
As a matter of fact, the fact that they showed a benefit makes it more likely that there is a benefit to it than not, they just cannot be sure. Think of a political poll, even if you are in the margin of error, the candidate ahead in the poll is more likely to be ahead than the person that was behind in the poll.
I said "effectively". A poll that shows a1% difference with a 5% margin of error is not showing a real difference. It is nearly just as likely that the 1% actually goes the other way.
You can NOT ignore the statistical significance of the result in drawing conclusions. If you form the question as a null hypothesis, the result does NOT allow you to conclude there is a benefit. It also, to be fair, does not allow you to conclude there is no benefit.
And yet your thread title says "showing that bid sniping does not work".
As has already been discussed ad nauseum here, bid sniping on nominally identical items that sell regularly and have a well-established value would not be expected to provide significant benefit. In that sense, the study confirms what is intuitively obvious.
EOM.
Would a more nuanced title have mattered?
Since you don't believe in statistics, it would have been wasted.
All comments reflect the opinion of the author, even when irrefutably accurate.
@CoinJunkie said:
The conclusion of the article is not that "sniping doesn't work":
In one auction of each pair we submitted an early bid, and in the other we submitted the same bid exactly ten seconds before its close. Our results, from a set of 70 such pairs, indicate no benefit to sniping: we found evidence of 2.54% lower prices for the sniped auctions, but we did not find this benefit to be statistically significant.
In fact, sniping results in slightly lower prices, which is a benefit in spite of what the authors state above. From my perspective, the only downside to sniping is that in a multiway (eBay style) auction, if you are the last person to enter a bid, you can potentially not win the auction even though your bid was the highest. That would be due to your bid not being at least a full increment higher than the current price at the time it was entered.
This is a completely ignorant statement. When the difference is not statistically significant, it means that there is effectively no difference as it is within the margin of error of the measurment.
It is interesting that you start your statement by accusing someone else of making a 'completely ignorant statement" then you go on to misstate statistics. If a study concludes that something is not statistically significant, it does not mean that there is effectively no difference. It means that the study was not able to conclude that their was a benefit, because of the limits of sampling. There is a big difference between that and saying their is no difference.
As a matter of fact, the fact that they showed a benefit makes it more likely that there is a benefit to it than not, they just cannot be sure. Think of a political poll, even if you are in the margin of error, the candidate ahead in the poll is more likely to be ahead than the person that was behind in the poll.
I said "effectively". A poll that shows a1% difference with a 5% margin of error is not showing a real difference. It is nearly just as likely that the 1% actually goes the other way.
You can NOT ignore the statistical significance of the result in drawing conclusions. If you form the question as a null hypothesis, the result does NOT allow you to conclude there is a benefit. It also, to be fair, does not allow you to conclude there is no benefit.
In your original statement you said the the study was showing their is no effective difference, but that is not what the study showed. The study was inconclusive. Which it looks like we agree on.
As for the polling example, assuming no bias in the poll while a 1% difference is not statistically significant. It is also true that if a candidate is leading in the actual vote, it is more likely that the poll will show the candidate leading than his or her opponent leading. In other words I would rather see my candidate ahead, even if it was within the margin of error than behind.
I had a snipe go off today where the winner sniped at 5 seconds and I sniped at 3 seconds. Their bid was 66 cents above mine with a $2.50 bid increment at that level. We were both 35% higher than next shown.
If I would have just maxed bid or bid at 7 seconds, their snipe would not have been enough for next increment and I would have won.
Comments
I agree, in the abstract. However, sniping wouldn't be expected to save much money with commonplace, easily valued items (as many have said). And as you noted, there are any number of flaws in the methodology of the study. And most importantly, I have made arguments above for why sniping cannot ever result in a higher final price to the winning bidder than had he/she bid the same amount at the beginning of the auction, but might result in a lower final price. Those arguments have not been refuted, although there has been a lot of babble about psychology, etc.
And that I agree with. But the two points aren't mutually exclusive.
ACTUALLY THEY HAVE BEEN REFUTED. You just refuse to acknowledge it.
Bit $100 on day 1. Underbidder won't go past $80. You are the first one to $80, you win.
You wait until the last day. Underbidder has already bid $80. You have to go one increment higher.
I can't make it any simpler than that. Even if EVERYONE bids exactly the same - no psychology or anything - you CAN pay more by sniping.
All comments reflect the opinion of the author, even when irrefutably accurate.
READ the conditions of the thought experiment!!! You make the same max bid either at the very beginning or at the very end of the auction. In the example you present, when you snipe, you LOSE the auction, you do not "have to go one increment higher". I've already pointed out multiple times that the downside to sniping is that you will occasionally lose when your max bid is equal to or higher than any other bidder's.
The logical conclusion is that if you prioritize winning the lot over getting the best possible price, sniping may not (although that is arguable) be a good choice. If getting the best possible price on any lot won is the priority, then sniping is DEMONSTRABLY (without running ANY simulations or empirical studies) your best strategy. The savings probably won't make you rich, but they are savings nonetheless.
Over and out!
That is not true. Dear God, read slower.
You bid $100 in both cases. You WIN In both cases. It costs you $80 when you bid day 1. It costs you $82 when you sniped.
CONTRARY to your claim that sniping can never cost you more.
Q.E.D.
All comments reflect the opinion of the author, even when irrefutably accurate.
I think the study is valid - but only for what they tested.
THEY TESTED WIDGETS! Items a population of a bazillion and that most people (including me) that wouldn't waste their time on a snipe bid because surely another widget will come along.
It is not a surprise that they came to the conclusion that they did.
Rerun the test with items that deserve sniping and I suspect you will get a far different result.
“In matters of style, swim with the current; in matters of principle, stand like a rock." - Thomas Jefferson
My digital cameo album 1950-64 Cameos - take a look!
It's hard to run the test with non-widgets. If you can't pair the "identical" coins, how do you know whether the resulting price difference is due to the timing of the bid or the coin itself?
I think the bigger problem with sniping on "unique" coins is that you have the possibility of two people going nuclear. It must have happened here or there, don't you think?
All comments reflect the opinion of the author, even when irrefutably accurate.
This is incorrect. The winning bid would be $81 either way. If you bid $100 first and then the underbidder won't go past $80 and bids that amount, ebay will automatically bump your bid to $81 if you have previously bid $100 as your high bid. That is how proxy bidding works. In the latter case, the underbidder bids the same $80 and with your snipe, you outbid his $80 and win the item at the next bid increment of $81.
Collecting 1970s Topps baseball wax, rack and cello packs, as well as PCGS graded Half Cents, Large Cents, Two Cent pieces and Three Cent Silver pieces.
Waiting for someone to yell "get off of my lawn"
Yes, you are correct with the bidding agent. You'd need a 3rd bidder for you to end up at $80. I make a friendly amendment. LOL.
You need a bidder at $79 and another bidder who refuses to go above $80.
All comments reflect the opinion of the author, even when irrefutably accurate.
I think I already did at one point.
All comments reflect the opinion of the author, even when irrefutably accurate.
Agree that it is harder to run the test with non-widgets. Frankly that is obvious. The closer you get to unique, the harder it is to make a valid comparison.
Nuclear bids are part of the game. I have been on both sides of a nuclear bid, but it depends on what your definition of nuclear bid is. If a nuclear bid is defined 50% over market, then I have lost most of my nuclear bids and won a few. But I never, ever bid more than my "walkaway" price. I win some and I lose some, and I don't get upset with a loss but I enjoy a win. However, others may bid way more than their "walkaway price" and either return the item or decide they want to be buried in it for a very long time. I've seen both as a seller and a buyer. I know it is frustrating for sellers sometimes, but that is the game we play.
“In matters of style, swim with the current; in matters of principle, stand like a rock." - Thomas Jefferson
My digital cameo album 1950-64 Cameos - take a look!
I agree. That's why I don't really worry about sniping and other games. I throw done my bid at my number. If I win, I win. If I lose, I lose. If I ever feel like I HAVE to win, I guarantee I'll end up over-paying.
All comments reflect the opinion of the author, even when irrefutably accurate.
Agreed. That is an edge case that would cost you more when sniping. It is unlikely enough and not particularly costly such that it doesn't change my belief that sniping saves the bidder money overall, but I will admit I missed it.
We agree.
Let's stop there.
MODERATORS, CLOSE THIS THREAD!
All comments reflect the opinion of the author, even when irrefutably accurate.
Glad you agree with me that sniping is a good idea overall. Now maybe you should modify the thread title.
Sniping is the only way I’ll ever bid and I’ve bid in 100’s of auctions on EBay. There are only 2 outcomes. I win the item for the amount I’m will to pay or less. Or I lose the item to someone willing to pay more. In live in person auctions when I am purchasing art for my collection I bid up to the amount I’m willing to pay. 2 out comes again. I win or I lose. Getting into a bidding war makes no sense whatsoever. You may win an item and have buyers remorse for way, way over paying.
"I spent 50% of my money on alcohol, women, and gambling. The other half I wasted.
If the early bid and snipe are the same amount they only matter in relation to the bids of the other bidders. So the question would be how do the bidding strategies effect the other bidder’s strategies.
With the snipe, the only possible impact on another bidder would be to guard against a snipe.
I guess I don’t see how sniping would lead to the other bidders bidding more than if the bid were placed early.
That’s making the assumption that the other bidder wouldn’t raise the bid if you bid earlier.
True, but he stated that up front.
This study has lots of issues - and there are a number of 'peer-reviewed' journals out there that publish poorly conducted 'science'.
Further, and they may have stated this but I did not notice because I gave up reading it, the conclusions only apply to the items they studied and in the time period they studied them. So extending their results to snipe/no snipe more broadly doesn't hold.
Etc.,etc....
Successful BST Transactions: erwindoc, VTchaser, moursund, robkool, RelicKING, Herb_T, Meltdown, ElmerFusterpuck, airplanenut
I will take the average 2,54% lower buy price.
They needed more data points than 70 to show 'statistical significance'.
I am not anti-science, but I am anti-poor interpretation of science.
I agree... I never thought of sniping as a way of saving money.
JUST WIN BABY
BHNC #248 … 140 and counting.
I did not read the study.
No way the conclusion is correct, no way no how.
I am a big proponent of science, but my gut says this study can not be accurate.
As a bid snipper for 20 years, I am sure I have saved ten of thousands of dollars.
How can one do such a study of unique items selling on eBay, impossible. Add end times, photo quality, and a million other factors.
The study may work if you want to buy a common coin widget.
And as Ricko said, (sometimes) it is about winning the item and not necessarily getting the best price.
I get out-sniped 50-75% of the time....almost always from other experts or dealers who know more than I do or who have deeper pockets or better avenues to dispose of their "kills" and/or "losers." It's all part of the equation. I'm happy getting something from time to time.
Bidding Early and winning....means you paid the most money of anyone in the country when all interested bidders had plenty of time to think about it. Sometimes that works and you get a fair deal, about the same as if you had sniped. But most of the time someone will take out your bid in the last minute....if only just to "take out your bid" and show you up....even if they have no intention of paying.
True story from last month.
Saw an Ebay item pop up with starting $99 bid with a $250 BIN. It was a scarce item and I wanted it. Most wouldn't have known what it was. I offered the seller $135 for it ($155 shipped and taxed)....sort of a "fair" low ball imo. I felt it was worth $175+.
Seller passed on my offer and wanted to see the item "run." We both agreed it could fetch considerably more than my offer. I check on the item with an hour to go and no bids....still at $99. Very strange. I couldn't explain why not a single bid for a quality, problem free item was not yet placed. Well, I got caught up in surfing that hour and didn't get back in time to put in a "nuclear bid." The item FAILED to sell. No bids. I was flabbergasted and upset that I missed out. It did end on Sunday midnight on the west coast....not a great time for an auction to end.
I got back to the seller and said I'd be willing to go a little over the starting bid since no one bid. We both must have missed something. They offered it to me around my original offer....at $130. I passed.....and said let's just let it run again and both of us take our chances. This time I was ready and placed my max bid of $135 with 5 seconds to go. Got it....for $99....or $115 all in shipped/taxed. No one but me bid on it. Once I got it in hand, I confirmed it was all there and worth in the $150-$175 range. For whatever reason, this one didn't catch anyone else's eye....there were dozen "watchers" on it though. Seller had reasonable FB and at 100%, with low shipping costs. It didn't help that the seller's description was a bit "jumbled" and could have been better.
I stopped at "ACADEMIC."
Good choice. I should have when choosing careers.
All comments reflect the opinion of the author, even when irrefutably accurate.
George Orwell, "There are some ideas so wrong that only a very intelligent person could believe in them."
.
If you try to enter that $500.01 bid in the last few seconds where there is already one in play at $500.00.....the Ebay system will not allow it to be entered. I've had numerous snipe bids rejected in the last couple of seconds because I wasn't a full bid increment higher than a max bid already entered and in play....entered a second before mine. Now if you bid $500.01 well before the bidding has escalated to the $500.00 level (say at $450), then the system may well allow a bid of $500.01 as the $500.00 bid is not in view yet. But I'm not sure of that. To be fair, that person holding the $500 bid should be able to also hold the entire bid increment just above it (ie. $500.00-$509.99). To allow it to be chiseled for $500.01 is not very fair.
Any arbitrary amount will be accepted by eBay as long as it's at least a full bid increment higher than the current bid. I believe Heritage is the same in terms of pre-auction Internet bids. Not sure I agree regarding fairness. Max bids are just proxies until they are actualized. If you want to "hold the entire bid increment", increase your max accordingly.
I'm not sure I understand what you're saying here, but if the current bid is $500, a bid of $501.01 will be rejected just the same as if you'd bid $500.01. If you've successfully entered either $500.01 or $501.01 as a max bid and someone bids $500 at the last second, you'll win in either case for your max.
I'm not sure why sniping or not sniping is so controversial.
Different people are more comfortable with different approaches and people should pick the bid strategy they are comfortable with.
It is interesting that you start your statement by accusing someone else of making a 'completely ignorant statement" then you go on to misstate statistics. If a study concludes that something is not statistically significant, it does not mean that there is effectively no difference. It means that the study was not able to conclude that their was a benefit, because of the limits of sampling. There is a big difference between that and saying their is no difference.
As a matter of fact, the fact that they showed a benefit makes it more likely that there is a benefit to it than not, they just cannot be sure. Think of a political poll, even if you are in the margin of error, the candidate ahead in the poll is more likely to be ahead than the person that was behind in the poll.
Join the fight against Minnesota's unjust coin dealer tax law.
The $501.01 snipe bid only goes off as a snipe if the prevailing high bid is at least one increment below the snipe amount (in this case no higher than $490.01), in which case the snipe of $500.01 will win the auction vs the $500.00 bidder. This is why it's always best not to enter bids at round numbers.
Collecting 1970s Topps baseball wax, rack and cello packs, as well as PCGS graded Half Cents, Large Cents, Two Cent pieces and Three Cent Silver pieces.
I generally agree that it's best not to enter (snipe) bids at round numbers, but neither $500.01 nor $501.01 is a round number. If the current bid is at $490.01 and that bidder's max is $500, and two last second snipers bid $500.01 and $501.01, respectively, whichever sniper bids first is the winner.
Good choice. I should have when choosing careers.> @Tomthecoinguy said:
I said "effectively". A poll that shows a1% difference with a 5% margin of error is not showing a real difference. It is nearly just as likely that the 1% actually goes the other way.
You can NOT ignore the statistical significance of the result in drawing conclusions. If you form the question as a null hypothesis, the result does NOT allow you to conclude there is a benefit. It also, to be fair, does not allow you to conclude there is no benefit.
All comments reflect the opinion of the author, even when irrefutably accurate.
Agreed. I was talking about a bid of $500.00 even.
Collecting 1970s Topps baseball wax, rack and cello packs, as well as PCGS graded Half Cents, Large Cents, Two Cent pieces and Three Cent Silver pieces.
And yet your thread title says "showing that bid sniping does not work".
As has already been discussed ad nauseum here, bid sniping on nominally identical items that sell regularly and have a well-established value would not be expected to provide significant benefit. In that sense, the study confirms what is intuitively obvious.
EOM.
All I can say is that sniping is my friend.
Would a more nuanced title have mattered?
Since you don't believe in statistics, it would have been wasted.
All comments reflect the opinion of the author, even when irrefutably accurate.
In your original statement you said the the study was showing their is no effective difference, but that is not what the study showed. The study was inconclusive. Which it looks like we agree on.
As for the polling example, assuming no bias in the poll while a 1% difference is not statistically significant. It is also true that if a candidate is leading in the actual vote, it is more likely that the poll will show the candidate leading than his or her opponent leading. In other words I would rather see my candidate ahead, even if it was within the margin of error than behind.
Join the fight against Minnesota's unjust coin dealer tax law.
I had a snipe go off today where the winner sniped at 5 seconds and I sniped at 3 seconds. Their bid was 66 cents above mine with a $2.50 bid increment at that level. We were both 35% higher than next shown.
If I would have just maxed bid or bid at 7 seconds, their snipe would not have been enough for next increment and I would have won.
I see sort by bid numbers is now back on US eBay.
Later.
got it thanks
OT
Phantombidder bit the dust.
Used them for years and years.
Any suggestions?
"Inspiration exists, but it has to find you working" Pablo Picasso