Question for skin2 about OPS
JoeBanzai
Posts: 11,804 ✭✭✭✭✭
in Sports Talk
I have stated in the past that I like OPS as a way to measure a batters value. At the same time I have had a few nagging problems with it as well.
Something skin2 mentioned in a post the other day got me thinking. He said that a walk was worth 2/3 (?) of a single, I am guessing that a HBP would also be worth less than a single. I don't know if there is a difference between BB and HBP, but for my question, I will assume they are the same.
Let's look at my favorite player Harmon Killebrew's MVP season; Harmon had a very nice OPS of 1.011. He walked and was hit by pitches 145 times that year. OBP counts those 145 as valuable as singles. If we adjust their actual(?) value down by 1/3, my figure on his OBP drops him down to .355 thus giving him an "adjusted" OPS of .939. Still a great number. Frank Howard's "adjusted" OPS for the same year puts him at .927 much closer to Killebrew's number than his .976 was to Killer's 1.011. I picked Howard because he walked about 40 times less than Killebrew, and had a much higher Batting Average.
I don't know if I should just put my calculator away and forget about it, or maybe I have just missed it being said before, but I would think that OPS rewards a guy who walks, say 100 times in a season with 150 hits and punishes a guy who gets 200 hits and walks only 50 times, if all the other numbers are the same. They would both have the same OPS, but the guy with 200 hits would be adding more offensive value.
Is my math incorrect? What am I missing here? Is it possible that I have a point?
Blaze away stat guys as well as skin.
Something skin2 mentioned in a post the other day got me thinking. He said that a walk was worth 2/3 (?) of a single, I am guessing that a HBP would also be worth less than a single. I don't know if there is a difference between BB and HBP, but for my question, I will assume they are the same.
Let's look at my favorite player Harmon Killebrew's MVP season; Harmon had a very nice OPS of 1.011. He walked and was hit by pitches 145 times that year. OBP counts those 145 as valuable as singles. If we adjust their actual(?) value down by 1/3, my figure on his OBP drops him down to .355 thus giving him an "adjusted" OPS of .939. Still a great number. Frank Howard's "adjusted" OPS for the same year puts him at .927 much closer to Killebrew's number than his .976 was to Killer's 1.011. I picked Howard because he walked about 40 times less than Killebrew, and had a much higher Batting Average.
I don't know if I should just put my calculator away and forget about it, or maybe I have just missed it being said before, but I would think that OPS rewards a guy who walks, say 100 times in a season with 150 hits and punishes a guy who gets 200 hits and walks only 50 times, if all the other numbers are the same. They would both have the same OPS, but the guy with 200 hits would be adding more offensive value.
Is my math incorrect? What am I missing here? Is it possible that I have a point?
Blaze away stat guys as well as skin.
2013,14 and 15 Certificate Award Winner Harmon Killebrew Master Set and Master Topps Set
0
Comments
WTB: 2001 Leaf Rookies & Stars Longevity: Ryan Jensen #/25
<< <i>Depends on what that guy did with the extra 50 hits. If his power was consistent through those hits, then SLG would stay the same and his OPS would get a boost by your fictional recalculation of walk and HBP values to OBP over the guy who walked more. >>
I specifically said all other hits were the same, only singles and walk numbers were changed, thus standard way of figuring OPS would be the exact same number. However IF a single is worth 30% more than a walk in offensive value, the guy getting more singles is getting his value diminished, while the guy who walks is getting unfairly rewarded.
If in fact a single is worth 30% more than a walk, and I wouldn't argue with that statement, that's a pretty LARGE number when comparing players offensive contributions.
Unless someone can explain where I went wrong with my math here, I for one am going to look at OPS differently than before, maybe not a lot, but I will not be giving the guys with the huge walk totals as much credit as those with higher BA when OPS numbers are similar.
Specific question to digicat; do you agree that a walk is worth 2/3 of a single in offensive value?
When I played ball we always used to say "a walk is as good as a hit", but I no longer feel that way.
As with any statistic, you can't just look at one number (in this case OPS) to get an understanding of a player's value. Heck, OPS doesn't even take into account their defensive benefit (or liability). WAR does a good job of putting as much value into one number as possible, but again, without context, every statistic is going to be difficult to gauge.
As far as your example of a guy who walks 100 times with 150 hits, as opposed to the guy who walks just 50 and gets 200 hits, I would also assume (you know what they say about assuming) the guy who walks more also strikes out less, as a strikeout is the least valuable out there is, whereas the guy who walks so much less is likely to have more strikeouts. I think it all balances out.
<< <i>When I played ball we always used to say "a walk is as good as a hit", but I no longer feel that way. >>
Same here.
On a walk, the batter only advances to first and any runners can only advance one base. By looking exclusively at hits, the ball is in play and you disregard outs, unless there's a base running blunder or comparable rare/unlikely play. You now have numerous possibilities (ie HR, double, clearing the bases, etc) and only limited by the number of runners on base.
Walks are good but the formula should give more credit to a hit.
<< <i>I have a hard time believing that a walk is only worth 2/3 of a single. As a batter, your one job is to get on base, and a walk does that as well as a hit. Dusty Baker aside, I think most every manager in baseball believes this. Throw in the fact that a patient hitter makes a pitcher throw more pitches, and I'd have to value the walk higher than just 2/3rds of a hit.
As with any statistic, you can't just look at one number (in this case OPS) to get an understanding of a player's value. Heck, OPS doesn't even take into account their defensive benefit (or liability). WAR does a good job of putting as much value into one number as possible, but again, without context, every statistic is going to be difficult to gauge.
As far as your example of a guy who walks 100 times with 150 hits, as opposed to the guy who walks just 50 and gets 200 hits, I would also assume (you know what they say about assuming) the guy who walks more also strikes out less, as a strikeout is the least valuable out there is, whereas the guy who walks so much less is likely to have more strikeouts. I think it all balances out. >>
First of all, I am not going to get pulled off topic here as happens so often. I am asking about OPS, not fielding or throwing or taking a lot of pitches. Just for once, I would like to get a straight answer to a SIMPLE question.
Your answer supports OPS as an accurate statistic, so if you are correct, that's fine. I am not really interested in dealing with strikeouts right now, again I would like to stay on point for a bit. But I will agree that negative value statistics have different ratings just as positive ones do.
If we can agree on ONE THING (probably impossible, but I'll give it a shot) a single has more value than a base on balls. The question I am asking is how much?
Here's what I could find on "Weighted On Base Average"; wOBA = (0.691×uBB + 0.722×HBP + 0.884×1B + 1.257×2B + 1.593×3B +
2.058×HR) / (AB + BB – IBB + SF + HBP).
Article on wOBA
Their claim is that a Unintentional Base on Balls is valued at .69 of a "point" but a single is not worth a full "point" but .88. Looks to me like their OPINION (?) is that a walk is worth 78% of a single. A bit different than skin2's mention of 2/3 value. I hope I am not misquoting you skin.
I would really like to hear from the "Stat-Heads" on this one. Does this kind of thinking give us a better idea of a hitters value? Seems to me it does.
<< <i>OPS does take into account extra base hits already, by factoring in slugging. So a guy who walks a lot is one who typically (not always, but usually) doesn't hit for a lot of power. Of course there are exceptions to this rule (Bonds comes to mind immediately), but by and large your leaders in walks are guys who don't hit for gobs of power. >>
"So a guy who walks a lot is one who typically (not always, but usually) doesn't hit for a lot of power."
WHAT? The top ten all time walk leaders are 80% sluggers with the exception of Ricky Henderson and Joe Morgan. Going down farther you still get mostly power hitters with a few exceptions, Yost, Rose and Collins. Fourteen out of the top twenty have over 400 homeruns.
You have got to be kidding me?!?!?!?!?!?!!?!?
<< <i>Specific question to digicat; do you agree that a walk is worth 2/3 of a single in offensive value? >>
No. But situationally I do know that a hit could be better than a BB. I don't know how I'd weight it though.
WTB: 2001 Leaf Rookies & Stars Longevity: Ryan Jensen #/25
The exact value can be estimated a variety of ways. Quick google search reminds us that linear weights says a single is worth 0.47 runs and a walk is worth 0.33 -- which I'm guessing is where the 2/3 comes from
But an out is worth -.27 runs, which would mean a single is 0.84 runs better than an out and a walk is 0.60 runs better than an out, which is in line with what wOBP says
Other issues worth noting:
For OPS it's necessary to remember than obp and slg are not equal. A team with a .300 obp and .500 slg will score far fewer runs than a team with a .400 obp and .400 slg. But for individual hitters context also matters. In a Juan Marichal-Sandy Koufax matchup in the 60s, a hitter who goes 1-4 with an extra base hit has done a lot to help his team. No matter how many times the hitter walks, he's still likely to be stranded, but a home run is a certain run, when only three or four runs in the whole game is likely. When games in Denver were averaging nearly 12 runs, going 0-0 with five walks was huge. Almost as good as having five singles, because any runners who would have advanced on those singles had a good chance of moving around anyways from the following hitters
I was curious in opinions on how much better. 1985fan doesn't think very much.
Your linear weight example states single = .47 of a run and walk =.33 would indicate a .14 difference that's pretty close, however your other example is .24 quite a bit more significant.
Would like to see a link to where you found the linear weight information. My quick Google search didn't find that one.
The rest of your post may have merit, but I am looking for individual numbers in order to compare similar players. There are many dynamics involved with a nine member team and how they mesh together. Good topic for it's own thread.
I would tend to think that so many more good things can happen as the result of a single than a walk because we are dealing with a batted ball, the higher number would be a better indicator.
http://www.tangotiger.net/RE9902event.html
Here is what is used for wOBP, note that it is always scaled to lgOBP:
http://www.beyondtheboxscore.com/2011/1/4/1912914/custom-woba-and-linear-weights-through-2010-baseball-databank-data
Actually, OPS tends to value the walk less. Here is why. Ops comprises two components, OB% and SLG% that are given equal weight. Pretty cut and dry.
In this equation, singles are counted in both OB% and SLG%, so they are counted double. While base on balls are only accounted for in the OB% portion(where they are given equal weight of a single).
So in OPS, walks are actually only about 50% of the value of singles.
However, more walks tend to occur with two outs, or a base open, so in those cases, they have much less value than the 2/3 of a single I said previously...more on that later!
Axtell may be the first person to actually over value a walk by stating it is just as good as a hit(single). Usually fans scoff at the notion of the value of a walk, even the 2/3 of a single value.
If you look logically at this, and you know appx half a batters plate appearances come with nobody on base, you will see that about half the time(that time being when nobody is on base), a walk and single are equal. So right away, half your walks and singles are 100% equal.
What knocks that value down to 2/3 is the cases where a single is obviously more advantageous than a walk. We all know those cases. A single with a man on third scores a run, while a walk doesn't. The walk still has value, but not the same as the single. A single with 2nd and 3rd is also more valuable than a walk with 2nd and 3rd. I think this is obvious to everyone. One can go on and on and see when a single is better.
However, the degree of how much better is where guessing is no longer a valid method. Sure, some people with a keen mind may be able to guess...but we don't have to guess, because we have millions of examples from the play by play data in MLB history to conclude to a very high degree how much each event is worth.
As stated above, where I said "more on that later." Not all walks in all situations are created equal. Actually, the same is true of any other hit too. A double with the bases loaded carries far more weight than a double with nobody on base, etc.
Since we have all this info in the play by play data, a run value(or run expectancy value) can be given to each event, and in each state of the combination of men on base and outs. There are 24 possible base/out combinations: Nobody one, zero outs. Nobody on, one out. Nobody on, two outs. Man on 1st, nobody out. Man on 1st, one out. Man on 2nd, one out. Man on 3rd, one out. Man on 1st/2nd, nobody out. Man on 1st/2nd, one out. You get the picture!
Each BB, 1B, 2B, 3B, HR, out made that occurs in all of those base/out situations has different values. The play by play data can see how much run expectancy occurs when each event happens in each situation.
Stats such as Situational Batter runs, Win Probablity added, and Base-outs added will give the most accurate example of the players value since it will put into context the base/out situation of each hitting event.
Here are some Base on Balls value examples: A walk with two outs and a man on third has the run value of appx .10 runs. While a walk with two outs and the bases loaded has the value of 1.06 runs. Quite a large difference in value, and obvious when you look at it logically. The math and real MLB play by play data simply solidifies the logic.
Actually, Win Probability Added also accounts for the inning and score too, taking it even a step further.
All one has to do is plug in each hitting event in each base/out situation and you get one heckuva clear picture on the value of that batter!
The standard Linear Weights that NVbaseball linked to is simply the average of each of those events. For example, instead of looking a the base on balls in each situations, it just takes the average of all situations and assigns its value that way, and it comes up with a .33 (depending on the year). It does the average for each event across all states. Some people don't believe in giving a double more credit because it happened with the bases loaded as opposed to nobody on. They just give it the average double and state that luck was responsible for the rest.
My biggest surprise in all of this, is that stats such as Win Probability Added are probably what every average fan wanted and believes in as it answers the 'clutch' questions, and the best "RBI man" questions, and gives one heckuva accurate picture on the value of the player that year...yet average fans scoff at them! That is what they should be embracing!
The Linear Weight traditionalists see value and ability as two different things. Value would be the Situational Base/out values, while ability would be the 'average' of the values of those. Choose your poison there I guess.
PS, As for WAR, the defensive equations and positional adjustments relies on voodoo and are not nearly as accurate as the stats from the play by play hitting data. Not even close.
In this equation, singles are counted in both OB% and SLG%, so they are counted double. While base on balls are only accounted for in the OB% portion(where they are given equal weight of a single).
So in OPS, walks are actually only about 50% of the value of singles."
I get that, but in order for both OB% and SLG% to be accurate by themselves, they have to be figured the way they are. You can't really figure walks into SLG% because you are measuring a players hits, just as you can't devalue walks in OB% because you are measuring simply that, on base %.
Your statement about walks being 50% of a single in OPS is then changed by saying that many walks actually have even LESS value than a single with men on base and 1st base open, that makes sense.
So if I have this right, OPS both under AND over values the walk, so does that mean your opinion is that it's effect is so small it "equals out in the wash" so to speak?
Everything else you say makes pretty good logic. The only thing I'll disagree with (nitpicking) is that a single with no outs and bases empty is going to have a SLIGHTLY higher value than a walk in the same situation because of the possibility of an error, but that's pretty rare.
After hearing your explanation of linear weighting, I am going to dismiss that as an accurate measure. I just couldn't get my head around the fact that each hit was worth .31 more than the last as you went from single to double to triple to home run. At the very least the home run has got to be worth more as it will always produce at least one run.
Very interesting information to look at. Thank you for the detailed answer. I have one last question regarding the walk, please excuse me if you covered it in your answer.
A walk to a player like Rickey Henderson, who leads off at least once a game, steals bases, a great base runner, plus next three batters should be the best in your lineup, has got to have much more offensive value, than a walk to my guy Harmon Killebrew, average to slow runner, never steals a base and the guys following him in the order tend NOT to be your better hitters).
Unless this is figured in one of the other rating formulas, I would think that unless we first group our players into at least 4 (leadoff, good overall hitters, sluggers, the guys who suck) categories no formula is going to work accurately.
Great stuff, nice to have a thread with a minimum of bickering and some really great information to study.
A few topics in your post. Without writing a book...a walk from the eight place hitter in the National League will of course have less value than a walk from a lead off man. THat of course is why you see a lot of 8th place hitters in the NL with high intentional walk totals. Of course, that is an extreme situation, and isn't very common throughout all lineup slots. I think you are getting at walks from a leadoff hitter compared to that of the best hitter, which would be a 3rd or 4th hitter.
The difference in value you seek is very minimal from those hitters.
Here is this years AL OPS from spot in the orders...
Leadoff batters .716
2nd .727
3rd .774
4th .783
5th .749
6th .738
7th .708
8th .682
9th .635
As you can see, the leadoff man has a .727 OPS guy behind him, followed by a .774. The third place hitter has a .783, followed by a .749. So a walk from a leadoff hitter, compared to a third place hitter really doesn't have much difference in value. Of course, the further you go down, the less it would have. But the elite hitters who draw walks usually bat third or fourth, and they are still followed by good hitters on average.
If a team has horrible hitters that are not taking advantage of an elite hitter who has a high OB%, then that doesn't reflect the value of the elite hitter, but rather the value and knowledge of the GM.
Also, if a team is batting a guy with a high OPS down in the order, and consequently not taking advantage of his ability, then that tells me the manager is stupid. It has nothing to do with the value of a walk, single, etc...
In general, sure, walks from the leadoff hitter carry a little more weight than walks from a cleanup hitter...but the difference is small, very small. In the overall scheme it really isn't worth looking at. Especially when you consider that home runs would have the opposite value, negating that value.
If anybody gets anything out of this, it should be the recognition that managers are stupid for batting guys at the top of the order on the basis of their contact rate, as opposed to their OB% ability...yet managers trot the post Colorado Neifi Perez guy in the two hole more than they should...which should be NEVER.
Two players with .950 OPS's are equal batters. Where or how a GM or manager uses them, or knows how to acquire good hitters, is a reflection on them, not the batter. Baserunning is another factor, and that can be measured independently.
The Linear weights is more accurate than OPS, so dismissing it doesn't make sense. Those figures are not guess, they are based on what actually happened in the millions of MLB play by play data.
For example, a home run with two outs and the bases loaded gives the HR a run value of 3.31, while a single in the same situation a run value of 1.53. An OUT in that situation is -.81 runs.
As you can see, the HR is not 4x the value of the single, and this is probably where you are getting stuck. You would probably think the HR is worth 4 runs and the single 2. That isn't correct, and we know this because of the play by play data and logic. First, four runs don't all go credit to the HR, some of that value goes to the guys who got on base.
Second, the single will still score two, advance the guy that was on first base, and the single itself still has a chance to score...which could equal the value of the HR. Sure, they don't always score, and that is why the value is only 1.53, compared to the 3.31. How do we know? The play by play data going back to the 1950's. Millions of examples.
When comparing players, if you and I both hit a hard double off the centerfield wall, we should get equal credit. Whether I had a fast guy on base that scored, and you didn't, should not come into the equation of judging me or you. Whether or not I had a good hitter behind me that drove me in, and you had a guy that stranded you, should also not come into the equation of judging me and you...and that is what the Linear Weights(both the average and base situations stats) eliminates, as it is fair. Whether or not a manager or GM knows how to make a lineup should not come into the equation of judging you or me. The Linear Weights gives the value based on the league average, context neutral situation, and rightfully so! That is what those figures above are showing...and they have MILLIONS of actual play by play MLB data to back them up! That is pretty strong evidence.
I don't agree it should be 1,2,3,4 like OPS or what linear weighting states .47*1B + .78*2B + 1.09*3B + 1.4*HR.
I like what wOBA says 0.691×uBB + 0.722×HBP + 0.884×1B + 1.257×2B + 1.593×3B +2.058×HR.
As you stated before, it's pick your poison.
<< <i>Thanks, Skin. That was a very informative and insightful read. >>
Thanks
Joe, most of those advanced stats come to a very similar conclusion in the end. Some have slight differences, and those do help spark good discussion when comparing players.
Like people say all the time, there is no way to get 100% to the right conclusion, but some devices do help you get close, and in the end, devices such as these measurements just make for a deeper understanding or affirmation of what a good instinctual participant kind of already knows or does.