June 24, 2008

WPA in the Gardenhire Era

Win Probability Added (WPA) measures how much impact specific plays had on the outcome of each game and assigns that value to the individual players responsible. For example, when evaluated by WPA within the context of each game hitting a grand slam in the fifth inning when the score is 10-2 has less value than drawing a walk in the eighth inning when the score is 5-5. Similarly, striking out to lead off a game is seen as less damaging than striking out down a run in the bottom of the ninth inning.

There are much better and longer explanations of WPA than that one, of course. If you're interested in learning more, Dave Studeman's WPA primer at The Hardball Times is a good place to start, and both Fan Graphs and Baseball-Reference.com offer tons of information on the subject. It's far from a perfect stat and isn't meant to predict how valuable a player will be or even definitely prove how valuable each player has been, but WPA is an interesting tool to use in looking back at what has already taken place.

In the past my looks at WPA have focused on current seasons, using the stat to break down monthly totals, examine player contributions within a particular stretch of games, or delve into how a specific game was won or lost. However, thanks to Fan Graphs recently releasing historical WPA numbers we can now use the data for bigger questions. For instance, how does WPA distribute the credit for the team's 574-473 (.548) record since Ron Gardenhire took over as manager in 2002?

Before tackling that, it's important to note than WPA doesn't measure defensive contributions, so strong defenders don't receive full credit for their value. Beyond that, WPA doesn't place offensive contributions in the context of position, so an .850 OPS from a catcher or shortstop is treated the same as an .850 OPS from a designated hitter or left fielder. Nothing can be done about measuring defense via WPA, but it's relatively easy to put the numbers in better context by using positional adjustments.

With the help of Fan Graphs creator David Appelman, I've taken the Twins' raw WPA totals from 2002 through the present and adjusted them based on the MLB average at each position. Most adjustments are minimal, but starters are given a boost relative to relievers and hitters who play up-the-middle positions are given a boost relative to hitters who man corner spots. The end result is an adjusted WPA that shows how much a player has contributed either below or above average for his position.

In six-plus seasons under Gardenhire, the Twins have used a total of 70 hitters and 60 pitchers. Many of those players failed to accumulate 100 plate appearances, so listing everyone would probably be a waste of space (even on the internet). Instead, let's break everyone down into three groups: 1,000-plus plate appearances, 400-999 plate appearances, and 100-399 plate appearances. First up, here are all the guys who either batted at least 1,000 times or faced at least 1,000 hitters under Gardenhire:

ADJUSTED WPA ABOVE AVERAGE: 1,000+ PLATE APPEARANCES

HITTERS adjWPA PITCHERS adjWPA
Joe Mauer 9.88 Johan Santana 21.97
Shannon Stewart 2.76 Joe Nathan 18.23
Corey Koskie 2.66 J.C. Romero 5.12
Jacque Jones 2.53 Brad Radke 4.02
Justin Morneau 1.97 Carlos Silva 3.25
A.J. Pierzynski 1.38 Juan Rincon 2.29
Torii Hunter 1.22 Scott Baker 2.28
Lew Ford 1.15 Kyle Lohse 1.60
Jason Bartlett -0.13 Matt Guerrier 1.17
Doug Mientkiewicz -0.74 Rick Reed 0.72
Michael Cuddyer -0.82 Boof Bonser -1.45
Luis Castillo -0.87 Joe Mays -4.83
Matthew LeCroy -1.16
Jason Kubel -1.46
Cristian Guzman -2.10
Luis Rivas -3.91
Nick Punto -6.96

Johan Santana and Joe Nathan blow away the rest of the competition, which perhaps shouldn't be much of a surprise. Santana emerged as an elite pitcher midway through Gardenhire's first season as manager, winning two Cy Young awards and two ERA titles while going 90-41 (.687) with a 2.92 ERA and 1,289 strikeouts over 1,179 innings. WPA pegs him as 21.97 wins better than an average pitcher from 2002-2007.

Nathan didn't join the Twins until Gardenhire's third year as manager, but has gone 19-8 with a 1.90 ERA and 388 strikeouts over 313 innings while converting 179-of-195 save chances for an amazing 91.8 percent success rate. WPA shows Nathan as 18.23 wins better than an average reliever, meaning that the Santana-Nathan combo was worth 40.2 games above .500 while the Twins as a whole were 101 games above .500.

Joe Mauer leads the hitters with 9.88 wins above average and his ranking significantly ahead of Justin Morneau shows the impact that positional adjustments have. Mauer (6.00) and Morneau (6.04) sport nearly identical raw WPA totals. However, given Mauer's playing time the average catcher produced -3.88 WPA, whereas given Morneau's playing time the average first baseman produced 4.07 WPA. Both players have been great, but compared to other catchers Mauer has been absolutely amazing.

Another spot where positional adjustments made a big impact is Torii Hunter. Many Twins fans would probably name Hunter as the team MVP during Gardenhire's time as manager, but his raw WPA was actually -0.18. That's not as bad as it looks, because zero WPA is equivalent to average and "average" equals a .500 record. In other words, according to WPA his offensive contribution was almost exactly average under Gardenhire. Still, Hunter being merely average is surprising.

He batted .274/.330/.480 with Gardenhire as manager, compared to the MLB average of .270/.335/.430. That puts him about five percent above average, but WPA docks him for grounding into a high number of double plays and having a poor stolen-base percentage while showing that he often failed to come through in crucial, game-changing spots. However, once you add in my positional adjustment, Hunter climbs to 1.22 wins better than the average center fielder even without his defensive value thrown in.

Joe Mays is the only pitcher with 1,000-plus plate appearances to post a truly horrendous WPA under Gardenhire, which makes sense given that his long-term contract kept him safe when most pitchers who performed as poorly as he did would have been dumped from the pitching staff. Nick Punto and Luis Rivas rate as the worst position players and combine for 10.87 wins below average, which makes sense given that Gardenhire loves middle infielders who can't hit.

ADJUSTED WPA ABOVE AVERAGE: 400-999 PLATE APPEARANCES

HITTERS adjWPA PITCHERS adjWPA
Bobby Kielty 2.04 LaTroy Hawkins 5.55
Mike Redmond 1.22 Pat Neshek 3.44
David Ortiz 0.24 Eddie Guardado 3.32
Jason Tyner -0.87 Francisco Liriano 2.28
Denny Hocking -1.44 Tony Fiore 1.48
Juan Castro -1.87 Jesse Crain 1.42
Luis Rodriguez -2.40 Dennys Reyes 1.04
Rondell White -2.46 Eric Milton 0.62
Dustan Mohr -2.54 Nick Blackburn 0.59
Kevin Slowey 0.46
Matt Garza -0.12
Ramon Ortiz -0.49
Livan Hernandez -0.97
Kenny Rogers -1.17
Terry Mulholland -2.04

Whereas the 1,000-plus plate appearances group was the place to find longtime lineup regulars and key starting pitchers, the 400-999 plate appearances group is home to lots of part-time players and relievers. LaTroy Hawkins, Pat Neshek, and Eddie Guardado were each fantastic out of the bullpen under Gardenhire--much like J.C. Romero in the previous grouping--while Bobby Kielty was so good in limited action that he had me freaking out when the Twins dealt him for Shannon Stewart.

Mike Redmond has been extremely productive considering his relatively minor role backing up Mauer, making for an amazing catching duo. David Ortiz was human in his lone season under Gardenhire, ranking 0.24 wins above an average designated hitter before racking up 26.2 WPA (and counting) over the next six seasons in Boston. Livan Hernandez has already posted the eighth-worst adjusted WPA among all Gardenhire pitchers, although Sidney Ponson's brief stint last season was still worse.

ADJUSTED WPA ABOVE AVERAGE: 100-399 PLATE APPEARANCES

HITTERS adjWPA PITCHERS adjWPA
Jose Offerman 1.10 Mike Jackson 1.27
Brian Buscher 0.05 Glen Perkins 0.39
Craig Monroe -0.14 Grant Balfour 0.27
Michael Restovich -0.14 Bob Wells 0.15
Brian Buchanan -0.26 Brian Bass 0.01
Chris Gomez -0.31 Joe Roa -0.03
Brendan Harris -0.53 Willie Eyre -0.19
Jeff Cirillo -0.71 Carmen Cali -0.34
Mike Lamb -0.75 Carlos Pulido -0.51
Tom Prince -0.80 Jack Cressend -0.51
Jay Canizaro -0.83 Matt Kinney -0.94
Michael Ryan -1.04 Sidney Ponson -1.02
Tony Batista -1.10 Seth Greisinger -1.03
Carlos Gomez -1.18 Aaron Fultz -1.22
Delmon Young -1.55
Henry Blanco -1.72
Terry Tiffee -1.74
Alexi Casilla -2.16

This group is where you'll find some young current Twins mixed in with one-season castoffs, mop-up men, pre-Redmond backup catchers, and failed prospects. Jose Offerman and Mike Jackson lead the way as the only significantly positive contributors in the 100-399 plate appearances bunch, as a 35-year-old Offerman batted .256/.363/.395 while coming up with several keys hit off the bench in 2004 and a 37-year-old Jackson posted a 3.27 ERA over 55 innings of relief in 2002.

At the other end of the spectrum, Alexi Casilla, Delmon Young, and Carlos Gomez have posted horrible WPA totals early in their Twins career, although Casilla has actually been slightly above average for a second baseman this year after being a complete mess last season. Oh, and the much-maligned Tony Batista managed to be 1.1 wins worse than an average third baseman despite playing just 50 games, and that's without factoring in his awful defense.

« Newer PostsOlder Posts »