As I've been promising for a couple of days, I'm finally coming up with a win probability graph for a Brewers game. (Thanks, again, to the SB Nation Mariner's site Lookout Landing for the idea.)
Win probability, as I've discussed a bit in the last few days, is a nifty stat that measures the--you guessed it--probability that a team wins at any given point in a game. At the beginning, it's just about 50% for both teams. As the game progresses, of course, as one team takes a lead and it gets closer to the end of the game (and thus, the number of opportunities for the trailing team to catch up goes down), the leading team's probability of winning gets higher and higher, and the losing team's probability continues to decrease.
It's a great way to measure the impact of relief pitching, as it doesn't rely on skewed stats such as earned runs, and it gives a pitcher more credit for working out of a jam than for getting the same number of outs with the bases empty. It's also handy to measure how relievers are being used--the closer the win probability to 50%, the more important the situation, so theoretically, the manager should be using his best relievers in those spots. As I compile Yost's usage data for relievers for the season thus far, it's distressing to see how many times Tommy Phelps came in when the game was on the line, just to save Turnbow for the 9th, by which point the Brewers already had a 95% chance of winning. More on that in a few days!
Yesterday was a major bummer, but if there is a silver lining, it's that the game made for a mighty interesting graph. When Deivi Cruz got on base in the top of the fourth, the Brewers chance of winning dropped below 20%. A mere half inning later, when Brady Clark drove in two runs to cap the 5-run rally, the chances were up above 90%. And, of course, it only took one more rally to send things in another direction. After Julio Santana retired Mike Matheny in the 8th inning, the Brewers chances were still at 78%. A single pitch--Ray Durham's pinch-hit double--sent those odds down to 24.1%. And, of course, it was all downhill from there.
Here, at long last, is the graph, showing the Crew's win probability from the first pitch (about 52% for the home team) to the final out:
- Best Brewers offensive performance: Brady Clark, +16.7%
- Biggest Brewers AB: Clark 1B off Reuter, 4th, +18.0%
- Biggest Brewers pitch: Santana strikeout of Matheny, 8th, +18.6%
- Worst Brewers AB: Weeks fly out off Walker, 9th, -8.5%
- Worst Brewers pitch: Durham 2B off Santana, 8th, -53.9%
And how 'bout that pitching staff (he says, with mixed sarcasm and, uh, sarcasm):
- Doug Davis, +9.4%
- Matt Wise, -36.5%
- Julio Santana, -43.9%
- Dana Eveland, -1.4% (this is one of the quirks of the system I'm using--Eveland did something good--got the final out of the 8th--but for some reason the win probability went down after he did it. I'm working on it...)
- Justin Lehr, +1.0%
This may be obvious, but to have two pitchers who each lessened a team's chance of winning by 35% or more (in the same inning, no less!) is pretty much a recipe for disaster.
Let me know what you think about all of this stuff. Would you like one of these for every game? (I'm not sure if I can promise that, but I may try...) What other stats would be of interest? Anything I can do to improve the presentation?