Parts I, II, III, IV, V, VI, VII, VIII, IX, X, XI, and XII

In the last entry we showed that Cy Young vote has correlated better to actual performance (i.e., Win Shares) than MVP in the last three and one-half decades. Now I would like to look at how pitchers have been rewarded by MVP votes.

In recent years a concept has seemed to take root, one that relievers, since they are "everyday players", should be considered for the MVP vote while starters should be relegated to the Cy Young. When I investigated the current concept that MVP candidates must be from a winner was more myth than reality. Does the same hold true for this aphorism?

To investigate this, I first had to look at the MVP vote distribution for all pitchers' Win Shares (min. 1 Pitching Win Share). I found that the lowest Win Share for a pitcher who received MVP attention was 5 (by the Browns' Elden Auker in 1941 along with 14-15 record and 5.50 ERA, 22% worse than the adjusted league average).

So I then culled all pitchers with at least 5 Win Shares in a season in which the MVP was awarded. I took the whole pitcher population and examined how well it correlated to the MVP. Here's what I found:

So it seems that this concept did take hold in the Eighties and has persisted since. And Willie Hernandez would like to thank the voters for it.

Next, I'd like to address this question from my friend Chris regarding my assessment that the "MVP must be from a winner" concept was more myth than reality:

My impression is that voters hold to the winners-only principle moreso on the top slot than the rest of the ballot. Didn't A-Rod win in 2003 essentially because the same people saying that by definition, he simply couldn't be the MVP, nevertheless listed him high on the ballot, while reserving their first place votes for loopy candidates like Shannon Stewart or David Ortiz? Maybe the modern MVP voter uses a hybrid approach.

Shannon Stewart?!? Did that really happen? It was not the MVP's proudest moment, but at least they didn't Mo-Vaughn the award to Ortiz or Stewart.

While it's true that Alex Rodriguez didn't have overwhelming support in 2003, he did get the most first place votes with six. Also receiving first-place votes were Carlos Delagado (5), Jorge Posada (5), Ortiz (4), Stewart (3), Manny Ramirez (1), Nomar Garciaparra (1), Vernon Wells (1), and Miguel Tejeda (1, it came in a year late). Here are the complete results. Also, A-Rod was the only man on all 28 ballots in 2003 (though he was ninth on one, surely a Seattle voter).

You'll recall that A-Rod lost in 1996 by the slimmest of margins (3 points) when he was Tildened by Juan Gonzalez. Rodriguez was one first-place vote behind Gonzalez (11 to 10), and of course, was stiffed by a couple of Seattle writers, who had to pay homage to Junior on their ballots before turning to the then-rookie.

Anyway, does the first-place vote reflect actual performance better or worse than do the entire voting results? Do voters consider the teams' performance more for first-place votes and then dole out the rest of the votes according to player performance?

Let's see….

First, what constitutes a player worthy a first-place vote? I ran the distribution for just those players receiving first-place votes, and found that the lowest Win Share total that merited such an honor was 12 (Harry "The Horse" Danning in 1937, the catcher for the NL Champ, the New York Giants, who played all of 93 games—quite un-horse-like). That will be our lower limit.

I ran the numbers for all players with at least 12 Win Shares. There's one other thing to keep in mind: there is no consistent record (at least not in Lahman or Baseball-Reference.com) of the first place votes in both leagues until 1937. I weighted the first-place votes against (multiplied by 10 in 1937 and 14 every year since) against the maximum number of points available.

Here are the results:

It appears that the voters have been extremely capricious over the years in awarding their first-place votes. I did worsen in the expansion era, though halfway through this decade, there seems to be some improvement.

So what is the cause? Do voters look more at team performance or are they just acting completely capriciously?

Let's compare team position and winning percentage for our group of ballplayers. Here are the results:

There is the slightest of correlations between getting first-place MVP votes and team performance, and that has been apparently declining over the years.

So what does this tell us? When it comes to first-place MVP votes, voters don't necessarily vote for the best candidate nor do they vote for the best candidate on a winning team. They vote for whomever they want. They ignore better players and players on better teams. And god bless 'em.