Trying to Make Peace with WAR

There’s been rich discussion in the baseball community about the ethics of the analytical revolution over the past several decades, accentuated and explored by the book/movie Moneyball.  Whether you agree with the philosophies of the statistic or not, Wins Above Replacement (or WAR) is the best publicly available assessment of a player’s total value.  On the contrary, in my opinion, runs batted in and pitcher wins are the more “established” and traditional statistics used to value players.

I looked at all qualified player-seasons from the Statcast era (2015-2019) and curated two scatter plots - one for pitchers and one for hitters.  To be fair to leadoff hitters, I’ve averaged runs scored and runs batted in. I compared the totals to fWAR, as shown below:

*Please note that this excludes all negative-WAR seasons

There is actually a strong correlation here!  The above chart demonstrates that sabermetricians and traditionalists really don’t have much to disagree on.  The r-squared value effectively states that 55% of run and RBI totals come from some contextual environment such as what runners are on base.  So we can be certain that about half of traditional stat totals come from how good a player actually is.

I’d argue that this is enough to the point where the two opposing factions of invested fans shouldn’t squabble over which stat is better - they say some similar things.  Though I much prefer WAR because it includes defense, baserunning, and is context-neutral, both measures are adequate ways to evaluate hitters offensively.

But what about pitchers?

*Please note that this excludes all negative-WAR seasons

Due to sample bias (only qualified seasons were included), the extreme winning percentages that would actually lower this correlation even more are excluded.  With that in mind, the correlation between pitcher wins and WAR is already a weak one. Jacob deGrom exemplified the discrepancy last year, when he won the NL Cy Young award^ with a 10-9 record but had a sample-leading 9.0 fWAR.

^I would argue that the media coverage of this discrepancy furthered the campaign for his victory

It’s hard for me to defend wins in this case as I did with RBI earlier.  They are so team and offense-dependent to become virtually meaningless. The correlation value gathered from the scatter plot above says that only 18% of a pitcher’s skill and on-field production has actually directly led to his win-loss record.

So, to recap: I believe it’s fair that the analytically-minded community takes issue with the pitcher Win stat due to its remarkably low correlation to performance, but people should calm the argument concerning runs and their batted in statistics because there isn’t as much to fret over.

Make peace, not war.  

But please, use WAR.

May your pennants fly forever.

-Ryan