MVP Votes and WAR Leaderboards, Converging Lists

Wins Above Replacement, or WAR, has become a pervasive statistic in baseball in recent years. The concept of evaluating a baseball player through the lens of wins, and in the context of his theoretical replacement, has taken hold in new spaces as well. WAR-type statistics have made their way into other sports and even into politics too. Baseball WAR comes in a couple of different iterations (FanGraphs’ and Baseball Reference’s formulations appear most regularly) and isn’t perfect (to some); however, WAR does an undeniably good job boiling down value nonetheless.

Regardless of how you might feel about WAR, there is a reason for its expanding presence. Having an effective catchall statistic to summarize value, or use as an outcome variable while projecting value, is a really powerful tool. As stakeholders in baseball have come to embrace what WAR offers, baseball has changed. Increasingly, a player’s WAR is leaned upon more readily than a handful of disparate statistics to get a snapshot of his effectiveness. This has far-reaching implications for baseball players’ careers and how the game is played. It also impacts how greatness is recognized.

To evaluate the changing relationship between player WAR and the recognition players receive for the value they present, I have drawn upon American League MVP votes from the last two decades. MVP voting is conducted via a sort of ranked choice system wherein voters rank their MVP top picks 1 through 10, with higher picks corresponding to more points. All points are added up to determine the number of cumulative points received, and thus, a ranking of all players who received any votes.

In this simple evaluation, two correlations were calculated. First, between each player’s (who received at least a single vote point) voting rank and their baseball reference (b)WAR. And second, between each player’s raw vote points (“Vote Pts”) received and their WAR. Those two correlations have been aggregated below. Because rank (1 being the highest rank, but lowest number) and WAR are likely to be negatively correlated, the absolute value of their correlation, in addition to the “Vote Pts.” correlation, make up the “Combined” value.

The low water mark presented above comes in 2004, when MVP rank/votes were only very loosely related to player WAR on the votes leaderboard. More specifically, there was nearly no relationship (0.05 correlation) between voting points and player WAR in the 2004 ballots. More recently though, voters’ cumulative ballots have far more closely corresponded to player WAR. In 2019, the correlations between first, player rank and player WAR and second, Vote Pts. and player WAR are quite strong ( >|0.75|).

One might imagine that, in MVP voting, pitchers are a bit of a confounding factor. While Justin Verlander is the only pitcher to win an AL MVP award in the last two decades, other pitchers got varying degrees of traction with voters. To check whether or not pitchers played an outsized role in determining these metrics, the next table excludes them from consideration.

By and large, it appears to be a wash. 2004 is still the low water mark for the WAR-votes relationship. More recently though (with pitchers aside), correlations have been stronger. In both tables presented, it looks like correlations spiked at the outset of the 2010s, following a half-decade of very weak relationships between WAR and vote ranking. Beginning in 2016 though, voters have adhered to WAR rankings far more closely.

These next two tables present a player-level view of the rankings for those seasons with the strongest and weakest vote/WAR correlations: 2019 and 2004. First, the 2004 AL MVP results:

There are several outliers on this list. For one, Michael Young received far more consideration than his WAR alone might dictate. In part, that is surely due to defensive metrics playing an undersized role back in 2004. That the 7+ WAR quintet of Miguel Tejada, Johan Santana, Ichiro Suzuki, Curt Schilling, and Alex Rodriguez failed to crack the top 4 in votes is pretty incredible. The top 5 in votes were, perhaps not incidentally, the top 5 in RBIs in 2004, which illustrates in part how much has changed in the voters’ preference for specific statistics. For Vladimir Guerrero, this was his only MVP award and it came in his first season as a member of the AL.

Finally, the 2019 AL MVP voting results:

The distribution of color in the 2019 MVP rankings looks far more like an even gradient. Still, Matt Olson hasn’t received the attention that he deserves, and a really strong season from Nelson Cruz as a DH still appears to get the benefit of the doubt against exceptional pitcher-seasons from Justin Verlander and Gerrit Cole.

The shortened 2020 season will surely represent an outlier season for player WAR and award voting, but it would be fair to expect this trend to continue given WAR’s heightened visibility today. While it is probably a positive development that voting more closely aligns to WAR, one does hope that the two leaderboards eventually just become a true reflection of one another.