<< <i>Wasn't computer grading tried once upon a time? >>
I know of five different attempts at computer grading. The esteemed Beth Deisher's out fit had involved a top technical research outfit and they came up with a working model. It passed from hand to hand and I believe is still around. There were four other attempts, possibly more. Like everything else, it is just a matter of time.
U.S. coin collector since 1943. Have tried desperatly to keep every coin that has come to hand, but unfortunately, some got away.
<< <i>Wasn't computer grading tried once upon a time? >>
I know of five different attempts at computer grading. The esteemed Beth Deisher's out fit had involved a top technical research outfit and they came up with a working model. It passed from hand to hand and I believe is still around. There were four other attempts, possibly more. Like everything else, it is just a matter of time. >>
And as a collector I won't pay any more attention to a computer generated grade than I do a TPG's now...in other words, none. The only grade on a coin that I am interested in is the one I give it myself...just like the only price I am interested in is the one I apply myself. If my price is either equal to or greater than the price the seller has applied then I am a buyer, if not, then I am not--irregardless of what some sheet says that I should be willing to pay.
Just because I'm old doesn't mean I don't love to look at a pretty bust.
Most of what people call eye appeal is quantifiable.
Someone will always have to stand over a computer's shoulder because they will consistently grade an increasingly smaller percentage of coins incorrectly. It's not so much the computer or the software that will revolutionize grading, it is the con- sistency.
It can also be the fact that coins can be graded on absolute standards rather than priced on relative ones.
<< <i>Someone will always have to stand over a computer's shoulder because they will consistently grade an increasingly smaller percentage of coins incorrectly. >>
Would you accept that the "person" standing over the computer's shoulder be a computer? And at what "increasingly smaller percentage" could we agree that we have achieved our goal? For example if the computer gets an "good job" (meaning, I agree) on 99.9999999999+% of it's grades, should we still quibble. If we do we kick it up a few pico-smacks. Maybe not today, but probably we can reach that point before too long.
However, ...
The intent of the original question ("Anyone interested in discussing how the price of coins is set?") posed that started this discussion was to see if others had any ideas. Personally I feel there is a lot of room for improvement, and to that end have had my company (software) working on a solution for the past several years. My goal is to try and help improve the hobby in general, and specifically to bring young people into the field. We have developed several software programs to manage several aspects of the hobby including collecting, buying, selling & trading as a collector, prices, and providing management tools for professional coin dealer's. We have many years in JiT, ERP, and many other disciplines that I feel would fit the coin business well, including sales & marketing automation. We have approximately 10,000 users, mainly collectors but a few dealers. If anyone would like to review the software (and the included several million prices, by the way), I will send them a CD - no charge. We have been collecting and compiling prices for seven years now. We currently update the price lists monthly but we could do it daily if there were a reason, like for example, real time for a dealer. My goal for this price list is FUD extermination, and consequently, lots more new collectors entering the hobby. Anyone interested can bash me here or contact me via eMail
U.S. coin collector since 1943. Have tried desperatly to keep every coin that has come to hand, but unfortunately, some got away.
Would you accept that the "person" standing over the computer's shoulder be a computer? And at what "increasingly smaller percentage" could we agree that we have achieved our goal? For example if the computer gets an "good job" (meaning, I agree) on 99.9999999999+% of it's grades, should we still quibble. If we do we kick it up a few pico-smacks. Maybe not today, but probably we can reach that point before too long.
However, ...
The intent of the original question ("Anyone interested in discussing how the price of coins is set?") posed that started this discussion was to see if others had any ideas. Personally I feel there is a lot of room for improvement, and to that end have had my company (software) working on a solution for the past several years. My goal is to try and help improve the hobby in general, and specifically to bring young people into the field. We have developed several software programs to manage several aspects of the hobby including collecting, buying, selling & trading as a collector, prices, and providing management tools for professional coin dealer's. We have many years in JiT, ERP, and many other disciplines that I feel would fit the coin business well, including sales & marketing automation. We have approximately 10,000 users, mainly collectors but a few dealers. If anyone would like to review the software (and the included several million prices, by the way), I will send them a CD - no charge. We have been collecting and compiling prices for seven years now. We currently update the price lists monthly but we could do it daily if there were a reason, like for example, real time for a dealer. My goal for this price list is FUD extermination, and consequently, lots more new collectors entering the hobby. Anyone interested can bash me here or contact me via eMail >>
I think this is great. There is a lot of room for improvement in price lists in the here and now and some of these changes should be easily implemented. Coins like a BU '83-P quarter wholesale at $25 but many price lists have them at $10 in choice or gem! Much of what holds the modern markets back is the crazy prices that are listed; Either the buyer has no interest in buying at list or the seller has no interest in selling at list. If a collector thinks a gem '83-P quarter is worth $10 he won't want to pay $30 for an example that isn't even in choice condition.
Right now computers could be developed that would get 75% of old coins and 95% of modern coins right on the money. Ironing out the obvious bugs in such a system could probably get it higher. I'd guess that if they can get about 98% then they could take over so it will probably happen in moderns first. When they hit about 99.9% then the definition of grade will probably just be whatever the computer says it is.
<< <i> I think this is great. There is a lot of room for improvement in price lists in the here and now and some of these changes should be easily implemented. Coins like a BU '83-P quarter wholesale at $25 but many price lists have them at $10 in choice or gem! Much of what holds the modern markets back is the crazy prices that are listed; Either the buyer has no interest in buying at list or the seller has no interest in selling at list. If a collector thinks a gem '83-P quarter is worth $10 he won't want to pay $30 for an example that isn't even in choice condition. /q.
BINGO! You've hit it right on. That is the intent and the idea. The methodology is that we've been collecting prices on every U.S. coin since 1616 from every source we can get them from. Dealers list's, ads, auctions, offer sheets, etc. We have derived price curves for each specific coin and as new prices are factored in they can redefine the curve. We also calculate a "weight" for each source which reduces the swings and the wild hares. We need about 2-300 good volume dealers to work with us as that would allow us to calculate both buy and sell prices anonymously in real-time. That would take about three to four months - depending on the number of transactions - to get the vast majority of the prices to very close tollerances. A friend that is one of the references in both red/gray complains often they never look at his bid lists and he is forced to buy at a lower price than he is willing to pay (also, those lists don't reflect real prices and that adversly affects his selling prices (lower than he wants)). We recognize and price approximately 1.3 million descreet coins providing 1 thru 70 (inclusive) in Mint State and 40 thru 70 in Proof, plus errors and varieties, attributes, and so on. We have a very fast db server that is chugging all day all nite and we are getting ready to upgrade to much faster multi-processor, 32mb memory and terabytes of hd. (sorry for the geek stuff - if you're not familiar with the jargon, not to worry, that just means it is fast.
U.S. coin collector since 1943. Have tried desperatly to keep every coin that has come to hand, but unfortunately, some got away.
Comments
<< <i>Wasn't computer grading tried once upon a time? >>
I know of five different attempts at computer grading. The esteemed Beth Deisher's out fit had involved a top technical research outfit and they came up with a working model. It passed from hand to hand and I believe is still around. There were four other attempts, possibly more. Like everything else, it is just a matter of time.
<< <i>
<< <i>Wasn't computer grading tried once upon a time? >>
I know of five different attempts at computer grading. The esteemed Beth Deisher's out fit had involved a top technical research outfit and they came up with a working model. It passed from hand to hand and I believe is still around. There were four other attempts, possibly more. Like everything else, it is just a matter of time. >>
And as a collector I won't pay any more attention to a computer generated grade than I do a TPG's now...in other words, none. The only grade on a coin that I am interested in is the one I give it myself...just like the only price I am interested in is the one I apply myself. If my price is either equal to or greater than the price the seller has applied then I am a buyer, if not, then I am not--irregardless of what some sheet says that I should be willing to pay.
Someone will always have to stand over a computer's shoulder because they will
consistently grade an increasingly smaller percentage of coins incorrectly. It's not
so much the computer or the software that will revolutionize grading, it is the con-
sistency.
It can also be the fact that coins can be graded on absolute standards rather than
priced on relative ones.
<< <i>Someone will always have to stand over a computer's shoulder because they will
consistently grade an increasingly smaller percentage of coins incorrectly. >>
Would you accept that the "person" standing over the computer's shoulder be a computer? And at what "increasingly smaller percentage" could we agree that we have achieved our goal? For example if the computer gets an "good job" (meaning, I agree) on 99.9999999999+% of it's grades, should we still quibble. If we do we kick it up a few pico-smacks. Maybe not today, but probably we can reach that point before too long.
However, ...
The intent of the original question ("Anyone interested in discussing how the price of coins is set?") posed that started this discussion was to see if others had any ideas. Personally I feel there is a lot of room for improvement, and to that end have had my company (software) working on a solution for the past several years. My goal is to try and help improve the hobby in general, and specifically to bring young people into the field. We have developed several software programs to manage several aspects of the hobby including collecting, buying, selling & trading as a collector, prices, and providing management tools for professional coin dealer's. We have many years in JiT, ERP, and many other disciplines that I feel would fit the coin business well, including sales & marketing automation. We have approximately 10,000 users, mainly collectors but a few dealers. If anyone would like to review the software (and the included several million prices, by the way), I will send them a CD - no charge. We have been collecting and compiling prices for seven years now. We currently update the price lists monthly but we could do it daily if there were a reason, like for example, real time for a dealer. My goal for this price list is FUD extermination, and consequently, lots more new collectors entering the hobby. Anyone interested can bash me here or contact me via eMail
<< <i>
Would you accept that the "person" standing over the computer's shoulder be a computer? And at what "increasingly smaller percentage" could we agree that we have achieved our goal? For example if the computer gets an "good job" (meaning, I agree) on 99.9999999999+% of it's grades, should we still quibble. If we do we kick it up a few pico-smacks. Maybe not today, but probably we can reach that point before too long.
However, ...
The intent of the original question ("Anyone interested in discussing how the price of coins is set?") posed that started this discussion was to see if others had any ideas. Personally I feel there is a lot of room for improvement, and to that end have had my company (software) working on a solution for the past several years. My goal is to try and help improve the hobby in general, and specifically to bring young people into the field. We have developed several software programs to manage several aspects of the hobby including collecting, buying, selling & trading as a collector, prices, and providing management tools for professional coin dealer's. We have many years in JiT, ERP, and many other disciplines that I feel would fit the coin business well, including sales & marketing automation. We have approximately 10,000 users, mainly collectors but a few dealers. If anyone would like to review the software (and the included several million prices, by the way), I will send them a CD - no charge. We have been collecting and compiling prices for seven years now. We currently update the price lists monthly but we could do it daily if there were a reason, like for example, real time for a dealer. My goal for this price list is FUD extermination, and consequently, lots more new collectors entering the hobby. Anyone interested can bash me here or contact me via eMail >>
I think this is great. There is a lot of room for improvement in price lists in the here
and now and some of these changes should be easily implemented. Coins like a BU
'83-P quarter wholesale at $25 but many price lists have them at $10 in choice or gem!
Much of what holds the modern markets back is the crazy prices that are listed; Either
the buyer has no interest in buying at list or the seller has no interest in selling at list.
If a collector thinks a gem '83-P quarter is worth $10 he won't want to pay $30 for an
example that isn't even in choice condition.
Right now computers could be developed that would get 75% of old coins and 95% of
modern coins right on the money. Ironing out the obvious bugs in such a system could
probably get it higher. I'd guess that if they can get about 98% then they could take
over so it will probably happen in moderns first. When they hit about 99.9% then the
definition of grade will probably just be whatever the computer says it is.
<< <i>
<< <i>
I think this is great. There is a lot of room for improvement in price lists in the here
and now and some of these changes should be easily implemented. Coins like a BU
'83-P quarter wholesale at $25 but many price lists have them at $10 in choice or gem!
Much of what holds the modern markets back is the crazy prices that are listed; Either
the buyer has no interest in buying at list or the seller has no interest in selling at list.
If a collector thinks a gem '83-P quarter is worth $10 he won't want to pay $30 for an
example that isn't even in choice condition. /q.
BINGO! You've hit it right on. That is the intent and the idea. The methodology is that we've been collecting prices on every U.S. coin since 1616 from every source we can get them from. Dealers list's, ads, auctions, offer sheets, etc. We have derived price curves for each specific coin and as new prices are factored in they can redefine the curve. We also calculate a "weight" for each source which reduces the swings and the wild hares. We need about 2-300 good volume dealers to work with us as that would allow us to calculate both buy and sell prices anonymously in real-time. That would take about three to four months - depending on the number of transactions - to get the vast majority of the prices to very close tollerances. A friend that is one of the references in both red/gray complains often they never look at his bid lists and he is forced to buy at a lower price than he is willing to pay (also, those lists don't reflect real prices and that adversly affects his selling prices (lower than he wants)). We recognize and price approximately 1.3 million descreet coins providing 1 thru 70 (inclusive) in Mint State and 40 thru 70 in Proof, plus errors and varieties, attributes, and so on. We have a very fast db server that is chugging all day all nite and we are getting ready to upgrade to much faster multi-processor, 32mb memory and terabytes of hd. (sorry for the geek stuff - if you're not familiar with the jargon, not to worry, that just means it is fast.