These days, OPS (On base percentage + slugging percentage) seems the be the best available stat to measure how good of a hitter somebody is. But I believe that we’re doing some double-counting, and that singles are given too much weight in the average. Couldn’t there be a stat that factors walks and HBP’s into slugging percentage that eliminates putting too much weight on ordinary singles. I mean, c’mon, every baseball fan knows that a walk’s as good as a hit.
On-base does consider walks.
Bill James was/is big on “runs created”:
RUNS= ((HITS+WALKS)*TOTAL BASES)/(AT BATS + WALKS)
Batting average IS counted twice in OPS (once in OBP, once in SLG%). I suppose if you subtract BA from OPS you might get a more useful statistic.
Baseball Prospectus is the place to go for statistical baseball analysis. They’ve a boatload of custom-tailored measurements (EQA, VORP, etc.) from which to choose.
A walk is only as good as a hit if nobody’s on base.
Example: There’s a runner on 2nd. If I get a hit, the runner is most likely going to make it to 3rd, possible score a run. If I walk, he’s still standing there on 2nd.
As to a better measure - who needs a better measure? Isn’t it far more fun to argue about which of two players is a better hitter than to just say “Smith is better, because his Snickelfish Score is 4 points higher”?
Well, a walk is not as good as a hit, for one thing. A single is a bit more valuable, because it advances runners without force and allows for runners to take extra bases.
Batting average is counted twice because it should be. A single, to use that example, has the element of both getting on base and advancing runners. If you counted it only once then you would be saying that a double and a strikeout is worth as much as two singles (or, in other words, that the value of the extra base on the double minus an out is worth more than a single.) But that is simply not true. Going 2 for 2 with two singles is much more valuable than going 1 for 2 with a double. Trust me; OPS as it’s used now is much better correlated with runs scored than it would be if you took out batting average. You can do the correlation analysis if you want.
There ARE better metrics than OPS. There are a lot of better metrics than OPS - RC/27, VORP, EQA, offensive winning percentage, even Win Shares, and you could whomp up some metrics that took OPS and modified it to make it more accurate. Heck, OPS+ is more accurate.
However, it takes a calculator and memorizing formulas AND knowing a boatload of league statistics to calculate those. No way a fan can add it up just looking at a scoreboard. OPS you can figure out just by adding two common stats together, so it’s the most convenient metric.
The double-counting of singles (and the first base of any XBH) is an annoying feature of OPS, and it’s the main reason I regard it as a cotton-candy stat rather than a serious one.
There are other problems with OPS, if the Oakland A’s brain trust is to be believed. According to Moneyball, they determined that each additional point of OBP was worth three times as much as every additional point of SLG. In the extreme, of course, it’s obvious - a team with a 1.000 OBP will score an infinite number of runs, but a team with a 1.000 SLG won’t. Apparently the disparity holds once you back away from the extreme.
The other thing about high-OBP players that Billy Beane points out is that, even when they’re not getting on base, they usually make the pitcher make many pitches in order to get them out. Since baseball is a war of attrition, and what’s being attrited is pitchers’ arms (Beane paraphrased), high-OBP batters help you use up the starter and get into the other team’s middle relief faster. And that’s almost always a good thing.
OPS is absolutely a serious stat and I don’t know why you would say otherwise. It is a crude stat, however. But it’s still extremely useful for day to day comparisons.
**
Don’t buy it. Just about everyone else in the stathead community except DePodesta believes it’s actually around 1.6. So OBP(1.6)+SLG. But again, that’s a pain in the butt so plain old OPS is very easy and useful.
OPS is a good stat, but there are better. It’s just they’re more of a pain to calculate.
OK, maybe I should rephrase that: it’s good for very fast, crude comparisons, but you wouldn’t use it in serious analysis.
The problem I have with this is, nothing’s really a pain to calculate anymore, unless you’re doing calculations in your head, or just doing one or two calculations of a single formula. Past that, you type your formula into a cell in your Excel or Quattro Pro spreadsheet, or into your TI-8x calculator, and let it do the heavy lifting from there.
So when I see OPS as a column in, say, baseball-reference.com’s hitting stats, I’m thinking, by that standard, it’s a cotton-candy stat. For the same amount of work, they could have done OBP + SLG - BA, or 1.6*OBP + SLG - BA, or whatever is most accurate without losing intelligibility. After all, just consider the BA double-count: two players with identical OPS could be as much as 100 points different on OBP + SLG - BA, and that’s not trivial.
Hell, no: (a) the burden of proof isn’t on me, and (b) running a correlation’s easy if you already have the data loaded up, but I don’t, and that’s the hard part. So forget it.
The theoretical problem with OPS double-counting singles, and double-counting the first base of every XBH, is that it effectively counts a walk as 1, a single as 2, a double as 3, a triple as 4, and a HR as 5. But it isn’t like that, because the biggest gap is between making an out and getting on base; the other differentials are all smaller. A walk isn’t quite as good as a hit, on average. But also on average, a walk is worth a hell of a lot more than 0.5 hits, and so forth. For instance, when would you rather have a triple followed by three outs, than a sequence of four walks? Never.
Every stat in baseball is deadly serious.
“Since 1950, Team A is 4-1 against Team B in the second half of a doubleheader when it’s a game rescheduled as the result of a delay and when they start a southpaw vs. a righthander.” 
So how would taking out BA help this? It wouldn’t. It would just make every single as good as a walk. Which isn’t true.
Nobody said OPS is perfect, but it’s damn good. 1.6OBP+SLG would be a little better, but OPS is just fine how it is. And it’s better than OPS-BA.
I still say that for the purpose of measuring the proficiency of a hitter, a walk is as good as a single. I agree that for the purpose of run production, a single is better than a walk. But I don’t think that that’s relevant to what we’re trying to measure.
If you’re not measuring the ability to produce RUNS, just what are you trying to measure? Producing runs is the purpose of hitting, is it not?
Depends. Which is closer to the relative value of walks, singles, doubles, triples, and HRs: 1,2,3,4,5, or 1,1,2,3,4? That’s what we’re disagreeing about.
I agree with RickJay that whichever stat correlates better with runs is the better measure of offensive value. Lacking a disc with spreadsheets full of MLB statistics, I’m in no position to run those correlations.
In the absence of such results, though, it seems pretty obvious to me that a single is, on average, worth closer to one walk than two, and there’s no alternative universe in which a homer and four outs is to be preferred to five walks and no outs.
If you’ve got evidence to the contrary, I’m ready to swallow my words, though.
Know what? You guys have convinced me that a hit is better than a walk, and that it should have more weight in the metric.
The way I was looking at it was: Whether Barry Bonds played for the Bad News Bears or the Giants, his prowess as a hitter would still be the same (discounting the reality that on the Bears, there would be no one “protecting” him, so he would draw a lot more walks). Therefore, it shouldn’t matter who’s on base. You always assume that no one’s on base. If this was the case, then a walk is as good as a single. But I was wrong. We really need to say that there are times that men will be on base, in which case a hit is more valuable than a single.
Alright, you guys seem to know what you’re talking about. So next question…
If you had to evaluate starting pitchers based on only 1 of 2 available stats (Win-Loss record or ERA), which one would you choose? We’re talking about pitchers that make 30-35 starts per year, so we can assume that they face comparable competition. I’d use ERA, but I have friend who thinks that W-L record is the better guage. We debate this all the time. What do you think?
Given only those two choices, I’d say that it’s ERA, and don’t see how you could argue otherwise. Pitching a full season on the Yankees as opposed to on the Tigers is pretty much a guaranteed 10 wins or losses, depending on which way you’re going. I mean, Mike Maroth might be 15-8 right now if he was the Yankees’ fifth starter. ERA doesn’t really strike me as a perfect indicator of a pitcher’s effectiveness, but at least it’s in the right direction. Given the right circumstances, a W/L record can be utterly meaningless, while a low ERA is pretty much always indicative of effective pitching.
Jimmy, my friend’s analogy is that when we measure how good a team is, everyone will agree that the team with the better record is the better team. Why don’t we use run differentila instead?
Also, when comparing 2 pitchers on the same team, you can assume that over the course of a season, they’ll get similar support. But the pitcher that wins 6-5 is the better pitcher than the one who loses 3-2. I don’t agree with that, but he’s a pretty smart guy so I want to know if anyone else agrees with him.
ERA without a doubt. W/L record is FAR too dependent on outside factor to effectively measure a pitchers worth.
JJ, that assumption you make there isn’t usually held up by the facts. Over the course of a single season it is VERY likely that two pitchers on the same team will receive radically different run support.
One of the little hidden truisms in baseball analysis is that a single season is too small to avoid sample size errors. Sad but true.
And as for the ‘run differential’ argument you’re exactly right. The ‘Pythagorean Record’ system compares RA and RS for an entire team to determine what their record SHOULD be and then you can compare that to their actual record to determine the element of luck over the season.
Here’s the formula:
Clay Davenport at BP has a more sophisticated ‘Pythagenport’ system as well.
Several thoughts, J. J..
First off, I don’t agree that the team with the better record is the better team because the teams don’t play the same schedules. When you get to, for example, feast on Detroit a lot, you W-L record is going to be inflated. I think run differential probably is a better indicator. It’s just not the one that matters, alas.
Secondly, the assumption that 2 pitchers on the same team will get similar support is just not true. It’s sensible, but it’s not necessarily true.
Finally, I agree with you; the pitcher who loses 3-2 is better than the pitcher who wins 6-5.
To take an extreme example, suppose we have a pitcher who pitches 35 games in a season, walks no one all season long, strikes out 700 batters, goes the distance in every game, and gives up exactly 35 hits all season. By any objective measure, he’s the greatest pitcher around.
But let’s say that those 35 hits were all homers, he got no run support all season long, and he ended the season 0-35. The fact that he got no run support all season is certainly not an indictment of his ability to pitch, and though he has a lousy record, he’s still obviously the best thing since sliced bread.
Obviously, this would never happen in real life. But the point is that the pitcher’s only job is to keep the other team from scoring runs, and people who value pitcher A over pitcher B because pitcher A is given better support are silly.
I see on preview that Johnathan Chance has said more or less everything I would have said had I thought of it, but I’m posting anyway, dammit.