Introduction

Based on the Sanders, Potter, Ehrlich, and Perline study on WAR and MLB MVP voting, the question is pretty straightforward: did MVP voters change after WAR became public around 2004? I think the answer is yes, but not in a perfect or automatic way. Voters already understood a lot of value before WAR, especially offense. The bigger shift is that voting started to line up more with total player value, especially the parts that were harder to see.

MVP Voting Explained

This chart shows how much more the voting lined up with measurable player value after WAR became public. Before 2004, the model explained 36.6% of the variation in MVP vote points. After 2004, that increased to 46.4%. That does not mean voters became perfect, but it does show their voting behavior became easier to explain statistically.

WAR Component Comparison

This chart shows where the voting changed. Offense was already rewarded before WAR, which makes sense because offense is obvious. The bigger change was defense. Defensive WAR was almost ignored before 2004, then became much more meaningful after WAR became public.

The after-WAR numbers are the original coefficient plus the post-2003 interaction, so the chart is showing the full before-and-after effect.

1. Regression Output

Regression Summary
Evidence Before WAR After WAR Explanation
Model fit 36.6% 46.4% The model explained more of the voting after WAR became public.
Vote points per WAR +35.24 +48.37 One extra WAR was tied to more expected MVP vote points after 2004.
Offensive WAR 0.569 0.639 Offense was already part of how voters judged value.
Defensive WAR 0.002 0.106 Defense went from basically ignored to part of the MVP case.
Pitching WAR 0.721 0.787 Pitching value mattered, but pitchers were still treated differently.

The regression output backs up the article’s main point. WAR was positive and significant, so better WAR seasons usually got more MVP vote points. The bigger result is the post-2003 interaction, which was also positive and significant. In other words, voting became more connected to WAR once voters had access to it.

The vote-point number makes that easier to read. Before 2004, one extra WAR was tied to about 35.24 expected MVP vote points. After 2004, that moved to about 48.37. So voters did not just start blindly picking the WAR leader, but the voting did line up more with total value.

The component numbers show where the change happened. Offense already mattered. Defense is the bigger shift, going from 0.002 before WAR to 0.106 after WAR. That is the clearest sign that WAR helped voters account for value that was always important but harder to see.

2. Voting Before and After WAR

Before WAR, voters still rewarded value, especially offensive value. That part is not surprising. Offense has always been easier to see and easier to sell in an MVP argument.

After WAR, the voting became more complete. The model explained more of the vote-point variation, moving from 36.6% to 46.4%. In normal terms, MVP voting lined up more with measurable player value after WAR became public.

The biggest change was defense. Defensive WAR was basically nothing before 2004 at 0.002. After WAR, it rose to 0.106. Defense did not suddenly become valuable. It was always valuable. WAR just made it harder for voters to ignore.

3. Pitchers vs. Position Players

Pitchers are the hardest part of this discussion because they are so different from position players. Baseball has one of the biggest comparison problems in sports. In basketball, soccer, hockey, tennis, golf, and lacrosse, positions are different, but they are still mostly playing the same basic game. Baseball is different. Comparing a starting pitcher to a catcher or shortstop is almost like comparing two separate jobs.

Football has a similar problem, but that is more of the exception. A quarterback, left tackle, cornerback, and edge rusher are all doing very different things. Baseball has that same kind of issue with pitchers and position players, especially because pitchers already have their own award.

Table 3 shows that pitching WAR was rewarded both before and after 2004, rising from 0.721 to 0.787. So voters were not ignoring pitcher value. The issue is that pitchers were still treated differently. In Table 2, the pitcher position coefficient is negative and statistically significant, which means pitchers were penalized compared to position players even after accounting for WAR and other controls.

That fits how people usually talk about the MVP. A lot of voters seem to view it as more of an everyday-player award because pitchers already have the Cy Young. Statistically, though, that still shows a bias.

Summary

WAR made MVP voting better, but it did not make it perfect. Voters became more connected to total value, and defense finally started to matter more. That is the strongest finding to me.

At the same time, WAR should not end the conversation. It is probably the best single-number tool baseball has, but MVP voting still has judgment built into it. The study shows WAR made voting more informed, not automatic.