Debunking the debunker
How many times have you seen it written or heard somebody say:
“Wins is a useless way to evaluate a starting pitcher”
Apparently at least one time less than it needs to be said.
The thrust of the post:
What we see is a very clear trend. As a pitcher’s ERA+ goes up (bigger values are better, 100 is average), their win total goes up. Are there exceptions? Of course. Every statistic has exceptions. But even in the face of contradictions, we still see a very strong correlation (r-squared = 0.51).
Um. Sure. Pitchers who have a better ERA tend to win more games. (Shocking, I know!)
But so what? I mean, it’s great that we have a statistical measure that can tell us Zack Greinke has been one of the ten-best pitchers in baseball this season. I guess. He’s sixth, three wins behind leader Adam Wainwright. Now, of course Wainwright has walked more batters, struck out fewer batters and given up more home runs than Grienke (and, for those worried about it, given up more earned runs as well) in spite of the fact that he gets to face a pitcher hitting in lieu of the DH for most of his games.
So how does Wainwright win more games than Greinke? Run support. The Cardinals have scored 5.51 runs per game while Wainwright is on the mound; the Royals have scored only 3.77 runs per game when Greinke pitches.
In other words, wins are a great stat for evaluating pitching, if you think that having Albert Pujols as your first baseman instead of Billy Butler is an important aspect of a pitcher’s performance. You can of course continue to use wins if you don’t need a pitching stat accurate enough to tell you that Carlos Zambrano has pitched better than Braden Looper this year.
In short, pitcher wins tend to work well in groups (if you look only at pitchers who qualify for end-of-season awards, yeah, the group of 15-win pitchers on the whole is going to be better than the group of 10-win pitchers). But they can be highly misleading if you want to use them to compare two individual players.
And for the kicker:
The problem with this post, is that taking a pro-Wins stance leads some to believe that we are anti-other stats. That couldn’t be further from the truth. Stats like ERA+, FIP and tRA are still better measures of how good a pitcher is (although we have minor quibbles with each). However, that does not mean Wins is a useless category. Nor does it mean there are 95 better ways to evaluate a pitcher.
In fact, in the absence of other stats, Wins is a very good, if not great, indicator of a pitcher’s value. So next time you hear somebody say Wins is a crappy way to evaluate a pitcher, throw a drink in their face and then make them read this post.
(Emphasis mine.)
But there isn’t an absence of other stats! Here, let me help you out:
There. Now you never have to worry about having no other pitching stats except for wins ever again. Problem solved.
Again, to review:
- Never use wins to determine a pitcher’s value, relative to other pitchers.
- Ever.
- I mean EVER.
Why in the world would someone write an article trying to suggest wins are a decent evaluative metric and then correlate ERA+ to win percentage?
Wins are a team statistic. End of discussion.
“And yes, in the absence of other stats, Wins still tells a good story.”
And when will there ever be an absence of other stats? Wins are useless for a pitcher because we have significantly better ways to measure the talent of pitchers. Why not use them? There will never be an absence of other stats so why talk about it?
“And yes, in the absence of other stats, Wins still tells a good story.”
WAYYYYYYY too much is being made about this one sentence. The point is just that Wins works (to an extent) as a stand-alone stat. I can look at a pitcher’s win total over time and make a general conclusion as to how good a pitcher he is. I would not NEED any other stats.
Now if I want to compare two pitchers with similar win totals, then yes, I will need other stats. But, in general, given enough of a sample size, Wins is a decent stat.
Over time you can look at a pitcher’s innings pitched and get an idea how good he is. IP has a stronger correlation to wins than ERA+ does. It’s about .85 to your .5. How come nobody talks about innings pitched being a great measure of how good a pitcher is?
Obviously the better pitchers will pitch more innings and therefore win more games.
Hits allowed also correlates better with wins than ERA+ does. The higher number of hits allowed, the more games you win. Nobody will ever say that allowing hits for a pitcher is a good way to measure their talent.
Cork – that might be true if the problems with pitcher wins as a measure of ability were randomly distributed. The issue is, they aren’t. If the 2009 season was expanded to 1,000 games, Grienke is going to continue to get less run support than Wainwright, because the Royals simply are a worse-hitting team than the Cardinals.
And next year? Probably the same thing, because the Cardinals have a number of good hitters that are expected to return, and the Royals have a number of poor hitters that are expected to return.
So no matter how big our sample, we cannot be assured that the issues with pitcher wins will come out in the wash.
I really don’t understand this Cork. What do wins add to the discussion? You don’t learn anything new from them about a pitcher. Obviously they are going to correlate well with ERA, because pitchers wins are an indirect by-product of pitcher performance; however, there is no point in using them when we have other, better stats.
If I was held a gun to my head to choose a pitcher to start game 7 of the world series, and the only info I had was the win totals for each pitcher; I would obviously choose the guy with more wins. However, I can’t think of another situation in which I would even look at wins.
The problem is that there are a lot of situations when looking at wins leads you to the wrong conclusion. Grienke has pitched much better than CC this year, yet he has fewer wins. He may lose out on the Cy Young for that.
“So no matter how big our sample, we cannot be assured that the issues with pitcher wins will come out in the wash.”
check the link below which is part 2 and looks at some individual pitchers. What you at saying is true to an extent. There are guys on the extremes of the distribution that don’t fit the model and it looks like offense is the biggest factor. I don’t have the data with me, but I did try to correlate with both some defensive stats and found very little correlation. When correlated with offense, there is an effect with the best and worst offensive teams. So only at the extremes.
In the link below, you will see that Cain has been the unluckiest. And the Giants offense is clearly to thank. With the filters I used, Greinke and Lincecum didn’t qualify. But greinke is on pace to be just as unlucky as Cain. And Lincecum is not far behind (~3 wins per year). But again, these are the extremes. I wouldn’t expect everybody to follow the pattern.
http://bit.ly/3rrbFw
Well said.
I laughed when they said “As a pitcher’s ERA+ goes up …their win total goes up” & used it as proof that win totals must show how good the pitchers performance was. In that same sentence, they just used ERA+ AS THE REASON why a pitcher’s win totals are good and claim win totals are an accurate measure! I’d like to hear them explain the win totals & ERA+‘s of Curt Schilling’s 2003 or Ben Sheets’s 2004 seasons. I’m sure that’d make for some laughs too.
“I mean, it’s great that we have a statistical measure that can tell us Zack Greinke has been one of the ten-best pitchers in baseball this season. I guess. He’s sixth, three wins behind leader Adam Wainwright. “
So what you are saying is that there are exceptions to the rule and that a statistic doesn’t explain everybody. (Shocking, I know!)
the point of the post was very simple. Wins is not a useless category. Is it the best? By no means. But it is not useless either. And IN GENERAL better pitchers win more games. Is that intuitive. Yes. But many seem to have forgotten this.
And yes, in the absence of other stats, Wins still tells a good story. In the absence of other outfielders, Jonny Damon is not all of the sudden a good outfielder. He still sucks.
I just don’t get why you’d use wins instead of something else. Winz is a replacement level stat and it’s so easy to find something else.
In what situation would you use wins? Sure, it correlates, but that doesn’t mean it has a use once you compare it to the correlation of other easily available stats.
No, there’s not one perfect stat. But to settle for an R^2 of .5 is weak. Winz tells us half of what we want to know? Between “knowing it all” and “knowing nothing” we’re settling for Winz at the half way point?
I dont remember anymore.
Who said anything about true talent? Citing Upton’s shoulder injury as the reason for his bad performance has nothing to do with this. He had a bad year—his wOBA reflects that.
Cain had a very good record of actual performance in 2007. But not only did wins fail to acknowledge that, they actually indicated the opposite.
and in comment #10 there is a link that addresses Cain as the pitcher whose win total deviates the most from his actual performance. On the bell curve, he is clearly one of the tails.
Speaking of Matt Cain…
In 2007 he had a 3.65 ERA, 3.78 FIP (10th best in the league), and 3.76 tRA. His K/BB ratio was over 2, and he allowed just .63 home runs per 9 innings (8th in the league).
He went 7-16, a .304 winning percentage. The Nationals this year had a .344 winning percentage. Why the hell would you use wins when you get absolutely ridiculous results.
In the presence of other stats, wins tells a horrible and misleading story.
BJ Upton has a .310 wOBA. In the absence of knowing that Upton has been battling a bum shoulder all year, wOBA would be a horrible and misleading indicator of how good a baseball player he is.
“BJ Upton has a .310 wOBA. In the absence of knowing that Upton has been battling a bum shoulder all year, wOBA would be a horrible and misleading indicator of how good a baseball player he is.”
But it’s a good indicator of how well he hit this year.
I guess what I’m missing in this whole thing is any sort of situation where I’d bother to use wins. MVP debate? Hall of Fame discussion? 1920’s players? When?
“But it’s a good indicator of how well he hit this year.”
and does that have any predictive value for next year or the year after?
“and does that have any predictive value for next year or the year after? “
Some. But we weren’t talking about predictive value of wins, were we? Just past value.
“a link that addresses Cain as the pitcher whose win total deviates the most from his actual performance.”
Implicit in this statement is that “win total” and “actual performance” are two different things.
Yes, I did imply that. I think at this point it’s pretty accepted that wins and performance are at least somewhat independent.