Recap
Box Score
Win Probability Graph @ Fangraphs
Highest WPA:
Paul Byrd .113
Jhonny Peralta .053
Casey Blake/Victor Martinez .040
Lowest WPA:
Fernando Cabrera -.213
David Dellucci -.133
Josh Barfield -.089
Since I've introduced WPA to these recaps, there's been many questions as to whether its values are really valid. They are, if you imagine the game in total isolation, and correct for defense and other observational context. WPA of a single game isn't going to tell you how good a player is, it tells you which players made the biggest difference in this particular game. If Casey Blake hits a grand slam with the Indians down three in the bottom of the ninth, WPA tells you that Casey essentially won the game. It doesn't care about his defense, or his batting average. In fact, it tells you nothing about Casey Blake other that he hit a grand slam, and that he was the driving force behind an Indians win. And in a single game recap, what better statistic to use than WPA?
To drive this home, let's take a look at the most important plays in the game, going in chronological order:
Top 1st Inning. Game Tied 0-0. Runners on First and Second, 1 Out. Indians' Win Expectancy: 53.6%.
Victor Martinez singles, making the score 1-0.
Win Exp. Increase: 8.9%
Top 1st Inning. Game 1-0 CLE. Runners on First and Second, 2 out. Indians' Win Expectancy: 58.5%.
Jhonny Peralta doubles, making the score 2-0.
Win Exp. Increase: 9.9%
Bottom 7th Inning. Game 2-0 CLE. Runner on First, 0 out. Indians' Win Expectancy: 69.8%.
Kendry Morales homers, making the score 2-2.
Win Exp. Decrease: 29.0%
Bottom 8th Inning. Game Tied 2-2. Bases Empty, 0 out. Indians' Win Expectancy: 38.7%.
Gary Matthews homers, making the score 3-2.
Win Exp. Decrease: 24.9%
So Paul Byrd, who spent most of the game increasing the chances of an Cleveland victory, ended the game with a positive WPA, while Fernando Cabrera's first pitch dropped the Indians' chances of winning almost 25%. It matters not who had better stuff, who looked better, or who's going to have the better season. In this case, it's not the process, but the result that matters. Of course, we who watched the game are going to illumate these cold hard percentages with context: pitching mechanics, what kind of stuff Byrd had, defensive plays, types of pitches, baserunning, managerial decisions, etc. And that's a good thing, because there are observations that we can make that WPA cannot.
WPA also provides a departure point for further analysis; for instance, it notes that David Dellucci, who I didn't really think about, had the second lowest WPA among Indians players. It turns out that he had at-bats during two very important junctures of the game, and in both cases made outs:
Top First. Game 2-0 CLE. Runners on Second and Third, 2 out. He grounds out.
Win Exp. Decrease: 4.6%
Top Ninth. Game 3-2 LAA. Bases Empty, 0 out. He flies out.
Win Exp. Decrease: 7.7%
If Dellucci singles in the first, the Indians lead 4-0, and perhaps knock Jered Weaver out early. If he singles in the ninth, it brings the go-ahead run to the plate with nobody out.
So I guess that while it's fair to label WPA as flawed as a standalone tool for analyzing a game, it's still a very useful guide if you provide the correct context.
Next Up: The series concludes. Sowers vs. Escobar, 3:35 PM.