Wins Suck

This past week's elections of Zack Greinke and Tim Lincecum as the Cy Young winners for their respective leagues has generated some controversy. While no one disputes the fact that these gentlemen are among the game's best pitchers, they fall short in one performance statistic that is near and dear to the hearts of baseball traditionalists: Wins.

Greinke was tied for seventh in the AL with 16 wins---three less than Felix Hernandez (second place among Cy Young voters), Justin Verlander (third place), and CC Sabathia (fourth place). In the NL, Lincecum was tied for fourth with eight other pitchers with 15 Wins---four less than Cy Young runner-up Chris Carpenter and two less than Adam Wainwright (third place).

The St. Louis Post-Dispatch's Bryan Burwell sums up the frustration of many writers.

I am not particularly outraged by any of this, but I am confused. Look, I think Lincecum is a heck of a pitcher, arguably the most gifted hurler in baseball. But I always thought the Cy Young was intended to honor the pitcher with the best season, not necessarily to reward the guy who has the best stuff....How can you look at what Wainwright did from a won-loss standpoint and essentially dismiss it in favor of Lincecum?....If won-loss records are suddenly obsolete, why do we bother to keep the stat?

The reason Wins were less relevant to some voters is that big-W Wins are not purely the product of pitcher performance. ESPN's Keith Law, whom Burwell singled out along with Baseball Prospectus's Will Carroll for not including Carpenter on their ballots, explained why pitchers aren't fully culpable for Wins and Losses with a question.

If a team got a $1mm bonus for winning a game, would you split it among all players who played, or give it to the pitcher who got the Win?

Winning a game requires many good things to happen, one of them is that the pitcher pitches well. But offense and fielding also contribute. A team that scores a lot runs or employs many slick fielders can pad a pitcher's Win total, while the opposite can result in undeserved Losses---the Royals were so bad in these other areas that the voters seemed to see the obviousness in this, giving him all but three first-place votes.

A recent analysis by Cork Gaines found that pitcher Wins explain about half of a team's run prevention---a fact he strangely used to defend Wins---and run prevention is only half of the game. In total, this translates to about 25% of a team's responsibility for wins. My own study finds that pitchers contribute even less than that. Now, it's easy to see why voters are beginning put less stock in Wins.

When choosing a metric for evaluating athletic performance, I rely on three criteria: 1) How much does it correlate with winning? 2) How well does it measure individual performance? and 3) How does it fit with our intuition?

Wins don't do so bad at identifying the impact on winning: Wins are correlated with run prevention and team wins. But, other metrics of pitcher performance that separate out the impact of offense (like ERA) are far superior for measuring how pitchers contribute to wins.

Wins do capture something about performance, but Win totals are polluted by fielding. While ERA is better than Wins, we need something more advanced to distinguish pitching from fielding. Voros McCracken showed that using strikeouts, walks, and homers (Defense Independent Pitching Statistics, or DIPS) to evaluate pitchers without their fielders does a better job of evaluating pitcher performance than ERA.

And finally, doesn't this all make sense? Who can't see the danger in equating pitcher Wins with team wins? This certainly isn't counter-intuitive; it's counter to the conventional wisdom---that's all.

So, kudos to Law and Carroll for using the right criteria for making their Cy Young picks. It's fine to disagree with their choices, but the reasoning behind their decisions is much more sound than the reasoning used the chorus of sportswriters who are condemning them.