April 24, 2006
WPA Through 18 Games
Yesterday's off day gave me a chance to update the Twins' Win Probability Added totals for this season. For those of you who are unfamiliar with it, WPA tracks what each play does to change the probability of a team winning a specific game. In other words, Torii Hunter hitting a grand slam in a blowout win over the Blue Jays wasn't worth nearly as much WPA as Lew Ford drawing a game-tying walk with two outs in the bottom of the ninth inning against the Angels.
On the surface Hunter's homer and the four RBIs that came along with it were more impressive, but in terms of actual impact on the game it pales in comparison to Ford's walk. Ford scratching and clawing his way to a free pass in a do-or-die spot against Francisco Rodriguez radically changed the Twins' chances of winning the game, while Hunter's grand slam basically just made things a little uglier for the Blue Jays in a game that was already pretty much in the bag.
The value of WPA is that it is able to account for both the event and the situation, and can spit out a value based on how things actually impacted wins and losses. It is far from a perfect stat and I'm certainly not comfortable relying on it to definitively decide who the best and worst players are, but it's certainly an interesting tool to have.
Before I get to the WPA totals through 18 games, I want to remind everyone that Will Young has been tracking WPA for every single game over at his blog. Will not only calculates the totals mere hours after the final out is recorded and posts them each night along with a handy graph showing how each team's odds of winning fluctuated throughout the game, he often includes an amusing bit of photo-shopping too.
It's not my intention to step on Will's toes, because I think what he's doing is great and I'll be linking to it constantly all season. In fact, his method for tracking WPA has additional value because he adds in a personal touch, adjusting the amount of credit given in special circumstances when a normal WPA calculation simply goes by what the play-by-play account says.
In other words, if Brad Radke gives up a 400-foot fly ball that Torii Hunter pulls back into the ballpark with a spectacular, homer-robbing catch at the wall, Will correctly gives Hunter the credit. Meanwhile, the by-the-book WPA calculation that I'll be tracking simply sees that Radke got an out in the situation and adjusts the Twins' chances of winning accordingly.
I've talked to Will about his methods for handing out WPA and I think it adds a ton of value to what is an excellent resource. However, I'm also curious about what the WPA totals look like without any biases or subjective adjustments thrown in. Think of Will's WPA tracking as a hot-fudge sundae and mine as a small bowl of vanilla ice cream. Most of the time the sundae is much better, but occasionally you just want the plain ice cream.
Grab a spoon ...
Joe Nathan 48.2% Luis Rodriguez -5.4%
Luis Castillo 44.1% Juan Castro -14.3%
Joe Mauer 40.0% Michael Cuddyer -15.2%
Justin Morneau 34.2% Matt Guerrier -19.3%
Shannon Stewart 31.4% Torii Hunter -20.1%
Lew Ford 30.4% Tony Batista -31.6%
Juan Rincon 28.7% Brad Radke -59.3%
Mike Redmond 27.5% Jesse Crain -66.1%
Francisco Liriano 20.0% Kyle Lohse -88.5%
Ruben Sierra 16.8% Carlos Silva -93.2%
Willie Eyre 7.0% Rondell White -125.3%
Johan Santana 4.0%
Scott Baker 3.4%
Nick Punto 1.4%
Jason Kubel 1.3%
The first thing that struck me about the above numbers is that 15 of the 26 players who have played for the Twins this year have posted a positive WPA total, yet because the Twins are four games below .500 the team total is -200.0. The reason for that is simple: Rondell White. White has been so bad thus far that his -125.3 WPA has nearly offset the combined contributions of the team's top three players.
Joe Nathan (48.2), Luis Castillo (44.2), and Joe Mauer (40.0) have each been worth slightly less than one win above .500, yet White has done nearly as much to drag the team below .500 all by himself. His .149/.157/.164 hitting line is not only horrendous, White has managed to fail in a remarkable number of important spots. Since WPA gives extra weight to at-bats that come in key moments, White's total suffers.
After White's jaw-droppingly low total, the next four least-valuable Twins have been pitchers. Carlos Silva's -93.2 WPA jives with his 8.33 ERA in four starts, and Kyle Lohse's -88.5 WPA matches up with his 11.57 ERA in three starts. Jesse Crain has been the worst of the relievers at -66.1 WPA, while Radke's -59.3 WPA ranking as just the third-worst total among the starters tells you all you need to know about the rotation's struggles.
Prior to the season I would have guessed that Tony Batista would be among the team's least-valuable players, but given his .279/.353/.459 hitting line thus far his -31.6 WPA is surprising. Batista has also come up with several memorable hits already and has two big WPA totals on his early resume--29.8 WPA in Game 7 and 25.1 WPA in Game 15--making his poor season total even more shocking. And remember that I didn't jury-rig these numbers one bit, so Batista earned them fair and square by coming up with games of -30.7, -21.0, -13.3, -12.8, -9.0, and -7.1 WPA.
Similarly, Juan Castro received a lot of praise for a hot start and some timely hits early on, but for the season his punchless .280/.308/.300 hitting line has dragged the offense down. Through 11 games Castro was among the team leaders in WPA at 31.2, but since then he's racked up a remarkable -45.5 WPA in seven games to push his season total down to -14.3 WPA. Once defense is properly factored in Castro becomes more valuable, but he's been far from the bright spot many fans and media members have made him out to be.
As a group the pitching staff has a cumulative -215.0 WPA and the hitters are at 15.0, which makes sense given that the pitching has been horrendous and the offense has simply been thoroughly mediocre. The few bright spots among pitchers have been Nathan and Francisco Liriano, who have each been nearly flawless, and Juan Rincon, whose one bad outing came when the Twins were already behind.
Among position players Castillo, Mauer, Justin Morneau, Shannon Stewart, Ford, and Mike Redmond have been very valuable, although nearly all of Morneau's value comes from the game-winning bloop single against Mariano Rivera in Game 11. Thanks to that hit Morneau had a 43.7 WPA for that game alone, which is the highest single-game WPA on the team. In terms of the impact on winning or losing a game, it's tough to top a game-winning hit off Rivera when your team is trailing with two outs in the bottom of the ninth inning.
Once you factor in some quick-and-dirty defensive value along with the WPA totals, the Twins' MVPs through 18 games are clearly Castillo and Mauer. That's not surprising given that Castillo has hit .404/.481/.468 with good defense at second base and Mauer has hit .316/.403/.421 with solid work behind the plate. Unfortunately, Mauer is a catcher who is not in the lineup every day and Castillo has already missed five games because of leg problems.
Nathan has also been fantastic, but thanks to Ron Gardenhire's hesitance to use him in non-save situations he's only gotten into five of the 18 games. The fact that Nathan has thrown 35 percent fewer innings than anyone else on the team and has been used about half as much as Rincon and Crain is simply poor strategy. Regardless of how bad the starting pitching has been and how few late leads there are to protect, a good manager would find a way to use his best reliever for more than five of the first 155 innings.