About a month ago, I did a
piece for The Hardball Times about a system that tries to predict who will win the MVP. I thought it might be interesting to look back at who the "MVP Tracker" predicts to win in previous years, and compare that with what actually happened. Today, I'll look at the AL voting over the past four years.
(Note: Don't directly compare the "PTS" between MVP Tracker and the actual voting; they use different scales. The thing to look at is the order, and even the size of the gaps.)
2003

This was A-Rod's last year with the Rangers, who finished 71-91, last in the AL West. MVP Tracker gets the top two right, but after that it's kind of a mess. Shannon Stewart, who was traded to the Twins on July 16, somehow conned his way into some votes with a VORP of 21 (This happened because the Twins went 46-23 after the trade; with Stewart hitting .322/.384/.470).
Manny, Nomar, and Ortiz probably split the vote to some extent. Although Manny pretty consistently does worse than expected because, well, he's Manny. Overall, MVP Predictor did horribly in '03, with a 0.23 correlation with the actual results among the top eight vote getters. This is hardly surprising, as the voting was wide open-
10 different guys got first place votes.
2004

MVP Tracker nails the Top 5 and the winner, but shuffles the order somewhat. I tried to incorporate an extra penalty for DHs into the system, but that never worked out; that's probably why Ortiz is too high. Again, Ramirez and Ortiz probably split votes, which allows Sheffield to jump up to 2nd. MVP Tracker did well in '04, with a 0.82 correlation among the top eight vote getters.
2005

This was the year when there was a big debate about whether Ortiz could win the MVP as a DH, and he came up short. MVP Tracker has A-Rod winning, with his advantage in BA making up for the deficit in RBIs.
My system is too high on Manny again. Please do not overlook his astounding UZR. And it really underestimates Vlad, who didn't have big power numbers, but had the "best hitter on a division winner" thing going on for him. Overall, a solid year for MVP Tracker, with a correlation of 0.73.
2006

I would first like to note that Travis Hafner, with a 79.7 VORP, was probably the best player in the AL last year. Just saying.
MVP Tracker had no doubt that Morneau would win, with all those RsBI on a division winner. (Side note: I feel like everybody, including myself, forgets that the Twins actually did win the Central last year, edging out the Tigers on the last day of the season. I guess that's because Detroit was in first all year, obviously went further in October, and who won the division didn't really matter all that much.)
The BBWAA just didn't do very well for themselves last year; Morneau and Thomas really have no place in this discussion. Ortiz and Hafner were probably the two best players, but, as DHs on non-playoff teams, were overlooked. Despite his defensive struggles, Jeter was pretty clearly deserving if you're going to give it to a guy on a playoff team.
2006 was MVP Tracker's best year, correctly predicting the top 3, and having a 0.83 correlation. Overall, I'm pretty happy with these results- it gets the winner right each time, and is right about the top two in three of the four years. And I'm pretty confident it will be right about this year's winner, not that that's particularly impressive.
In the THT article, I write about how my system thinks Holliday will win this year. If the Rockies had missed out on the playoffs, it would have predicted Rollins. I mention this because it's unclear to me when the MVP votes are due in. I've heard that it's the Monday after the regular season ends, even though there was technically one final regular season game on Monday night. If this is true, and the BBWAA doesn't take the 163rd game into account, I think Rollins wins.