October 3, 2007

My Team MVP Ballot

With the recently completed season still fresh in everyone's minds, I figured now would be a good time to examine who the most valuable (and least valuable) Twins were this year. There are certainly any number of different ways to approach such a task, but my preferred method is to utilize several different player-evaluation metrics and attempt to synthesize the information from them to form what should be one well-rounded opinion.

It'd be easy to go by gut feel, conventional wisdom, or a single statistic, but taking a wisdom-of-crowds approach by involving as much objective information and evidence as possible is more worthwhile. Rather than post a bunch of numbers and reveal my team MVP ballot, let's walk through the different metrics that I used and examine what each one had to say about assigning value to Twins players this season. If you're unfamiliar with something being referenced, click on the links to learn more.

First up is Value Over Replacement Player, which takes a player's production and compares it to the "replacement level" for his position, which is a level of performance that's considered readily available and easily obtained. In other words, how many runs was someone worth compared to a low-level player who a team could acquire at minimal cost. Justin Morneau gets compared to first basemen, Torii Hunter gets compared to center fielders, and so on, with playing time being a big factor.

                    VORP                            VORP
Johan Santana 60.0 Jason Tyner -0.5
Torii Hunter 39.2 Sidney Ponson -5.8
Matt Guerrier 36.8 Luis Rodriguez -7.7
Carlos Silva 34.6 Alexi Casilla -10.9
Joe Nathan 32.9 Nick Punto -27.1
Joe Mauer 30.2
Justin Morneau 28.8
Scott Baker 24.9
Pat Neshek 21.9
Michael Cuddyer 18.4
Jason Bartlett 14.7
Jason Kubel 13.5
Matt Garza 10.8
Luis Castillo 8.2
Boof Bonser 6.5
Mike Redmond 6.4
Ramon Ortiz 5.9
Kevin Slowey 5.4
Jeff Cirillo 1.0
Juan Rincon 0.7

As you can see, because replacement level is such a low baseline for comparison, few players racked up negative totals. Despite that, Nick Punto managed to be 27.1 runs worse than a replacement-level player offensively by hitting .210/.291/.271 in 536 plate appearances. Not only did -27.1 VORP rank dead last in all of baseball, it's the fourth-worst total any position player has posted since 1959, topping only George Wright (-33.1 in 1985), Neifi Perez (-28.0 in 2002), and David McCarty (-27.3 in 1993).

In terms of production and playing time, the season Punto just turned in offensively was among the worst handful of the past 50 years. At the other end of the spectrum, Johan Santana leads the team by a wide margin with 60.0 VORP. Santana was worth 57.3 runs more than a replacement-level pitcher and tacked on 2.7 VORP by hitting .286/.375/.714 during inter-league play. Hunter leads all position players with 39.2 VORP, which doesn't include his defensive contributions in center field.

Next up is Linear Weights (with a positional adjustment added), which is somewhat similar to VORP except that all comparisons are made to "average" rather than "replacement level." Average is a much higher baseline for comparison, so the result is far fewer positive contributions and far more negative contributions. In other words, compared to replacement level nearly everyone except Punto comes out looking decent. Compared to average, a lot fewer players keep their head above water.

                    LWTS                            LWTS
Johan Santana 33.1 Jason Kubel -0.3
Matt Guerrier 25.0 Mike Redmond -0.9
Joe Nathan 24.1 Michael Cuddyer -1.0
Joe Mauer 14.3 Jason Bartlett -2.2
Pat Neshek 13.4 Kevin Slowey -2.6
Carlos Silva 10.4 Ramon Ortiz -5.0
Torii Hunter 8.4 Jeff Cirillo -5.5
Scott Baker 7.7 Juan Rincon -5.5
Justin Morneau 1.2 Luis Rodriguez -10.4
Matt Garza 1.0 Sidney Ponson -10.5
Luis Castillo 1.0 Jason Tyner -12.6
Alexi Casilla -13.4
Boof Bonser -14.0
Nick Punto -30.8

Santana once again leads the team by a wide margin, while Punto again brings up the rear with an awful total. Morneau is a good example of the difference between VORP and Linear Weights, as his hitting goes from being worth 28.8 runs above replacement level to just 1.2 runs above average. While the caliber of readily available first basemen is clearly inferior to Morneau, the average MLB first baseman batted .276/.357/.463 this season compared to Morneau hitting .271/.343/.492.

Compared to other players at his position, Morneau had a slightly below average on-base percentage and a slightly above average slugging percentage, which adds up to essentially being an average hitter overall. Hunter fares better than Morneau in Linear Weights, but is no longer the top-ranked position player thanks to the average MLB center fielder batting .272/.338/.420 compared to his .287/.334/.505. An 85-point edge in SLG is big, but a below average OBP keeps him from racking up a huge total.

By comparison, Joe Mauer is helped by switching from VORP to Linear Weights and vaults ahead of Hunter as the top position player because his .293/.382/.426 hitting line was far superior to the average MLB catcher batting .256/.318/.394. Along with the changes involving Morneau, Hunter, and Mauer, you'll notice that Matt Guerrier ranks ahead of Joe Nathan in both VORP and Linear Weights, which is explained in part by both systems viewing all innings and plate appearances as equal.

In other words, tossing a scoreless inning with a 10-run lead counts the same as tossing a scoreless inning in a tie game. The third and final metric differs from that, as Win Probability Added is designed specifically to measure the impact that each play has on a team's chances of winning. While VORP and Linear Weights place the same value on a home run whether it comes in a tight game or a blowout, WPA values a walk-off homer much differently than a solo shot in a 15-2 game.

That's an important distinction when it comes to Guerrier versus Nathan, because Nathan typically pitches in higher-leverage situations. Thus, on average WPA views his scoreless innings as being more valuable than Guerrier's scoreless innings (and views Nathan giving up a run as being more costly than Guerrier giving up a run). As with Linear Weights, WPA uses average as a baseline and I've added positional adjustments because they aren't built in.

                     WPA                             WPA
Joe Nathan 3.15 Mike Redmond -0.04
Johan Santana 2.81 Michael Cuddyer -0.44
Carlos Silva 2.43 Ramon Ortiz -0.45
Pat Neshek 2.36 Boof Bonser -0.50
Scott Baker 1.45 Justin Morneau -0.64
Joe Mauer 1.03 Jeff Cirillo -0.73
Matt Guerrier 1.02 Juan Rincon -0.86
Torii Hunter 0.74 Sidney Ponson -0.92
Matt Garza 0.49 Jason Kubel -1.05
Kevin Slowey 0.42 Jason Tyner -1.41
Luis Castillo 0.24 Luis Rodriguez -1.64
Jason Bartlett 0.17 Alexi Casilla -1.99
Nick Punto -2.78

As you can see, Nathan benefits greatly from WPA because a) he was fantastic this season and b) the vast majority of his appearances came in tight situations. If you perform well in key spots, WPA rewards you. If you struggle in key spots, your WPA total plummets. All of which is why the WPA rankings look quite different than the rankings for VORP and Linear Weights. VORP and Linear Weights look at how you performed overall, while WPA looks at how you performed in specific situations.

One thing that all three metrics have in common is that they fail to account for defensive contributions, which is obviously a major factor when assessing a player's overall value and a big part of why all three rankings are pitching-heavy at the top. To rectify that problem, I've gathered fielding stats from The Hardball Times, Ultimate Zone Rating, and Baseball Prospectus, weighing them equally to come up with a three-headed defensive value for each player.

Similarly, I combined VORP, Linear Weights, and WPA to come up with three-headed values for hitters and pitchers. Because the baselines are different and some metrics deal in runs while others deal in wins, it's impossible to combine them to get one "big number." Instead, I've tried to weigh everything equally so that each metric accounts for one slice of the overall pie. By combining VORP, Linear Weights, and WPA with three different defensive tools, here's what my ballot for team MVP looks like:

 1. Johan Santana             14. Matt Garza
2. Joe Nathan 15. Luis Castillo
3. Joe Mauer 16. Kevin Slowey
4. Torii Hunter 17. Jason Tyner
5. Carlos Silva 18. Jeff Cirillo
6. Matt Guerrier 19. Ramon Ortiz
7. Pat Neshek 20. Juan Rincon
8. Scott Baker 21. Boof Bonser
9. Jason Bartlett 22. Sidney Ponson
10. Justin Morneau 23. Luis Rodriguez
11. Michael Cuddyer 24. Alexi Casilla
12. Mike Redmond 25. Nick Punto
13. Jason Kubel

The Twins Cities chapter of the Baseball Writers Association of America predictably voted Hunter the team MVP, but the numbers don't really support that in large part because the various fielding metrics no longer view him as an elite defensive center fielder. There's certainly a good argument to be made for Hunter being the most-valuable position player and I may have even made it prior to examining the numbers, but I think it's pretty clear that Santana was the Twins' MVP in 2007. Again.

Once you're done here, check out my latest "Daily Dose" column over at Rotoworld.

No Comments

No comments yet.

Sorry, the comment form is closed at this time.