I wrote a program to take play-by-play data from Yahoo Sports and transform it into win expectancy calculations. Combined with Excel, we can make cool pictures so you can see roughly how much players are helping us win. I've downloaded all the A's games so far and I've analyzed the data through June, so this little introduction will be about the first half of the season.
The basic idea is that for any stage of a game, if you know the inning, outs, baserunners and score difference, you can look up how teams in the past did under those circumstances. This is a brute force way to estimate the probability you will win. By seeing how a player affects their team's chances of winning, you can see how important their contributions really are. So for instance, a homer in a blowout counts for almost nothing while a single that brings you from behind into the lead counts a huge amount. Similar ideas apply to pitchers: striking out the side for a three-run lead save is basically worthless, but in a tie game, 9th inning with a runner on third, getting a pop-up for the second out is huge. The program automatically dings players for errors, but I didn't go back to adjust credit for spectacular defensive plays (it's possible... I just didn't have energy).
I can explain more details in some other diary. On to some pictures. This may be more fun for you if you read the text before the pictures, because there are some mini-quizzes for you.
For instance, here is a picture of Chavez's first-half.
No surprise, huh? Every picture in this diary is a running total of the player's contributions to the season, so everywhere the graph goes up is where he's making positive contributions. You can see he helped the team in mid-May and then tailed off before picking it back up on the first day of June and going on a long positive tear.
What does Chavy's performance look like in context? ....
Here is a graph of two hitters who contributed about as much as Chavy for the first half, along with one ROCK hitter and one erratic player. QUIZ. Can you guess who the other four players are?
Okay, Kotsay and Kendall got off to fast starts but stopped contributing (though they didn't hurt the team either). The erratic player was, of course, Byrnes, and the ROCK was, unbelievably to me, Kielty. He was the offensive foundation of the team for the first half. He's kind of slumping recently so I forgot, but I gotta give him credit for all his first-half contributions.
Once again note that we punish for errors but haven't (yet) rewarded for great fielding.
Okay, two more quizzes before I get too tired. Here are the 6 starters from the first half (not including Etherton). Can you figure out who is who without peeking?
The surprising thing to me is that, (1) by this measure, Saarloos contributed as much as Zito in the first half, though neither one has a lot to brag about; this probably means they both lived and died with their run support and didn't do a lot in aggregate to overcome bad run support or sabotage good support; (2) Harden is so kickass it's astounding.
This picture shows members of our bullpen. By the end of the first half, you can see three of them made positive, increasing contributions, two of them neither helped nor hurt, and two of them quite dramatically plummet off cliffs. Name them.
The possibly surprising thing here is that the much-trod-upon Rincon is only a small liability in the end. His blown opportunities are mostly made up by his coming through in high-leverage situations. One can also see that the middle of May was a very very bad time for the bullpen.
Okay. Getting tired. I will probably do at least one more of these diaries, with a graph about the 2B competition, a graph about the injured folks and the call-ups, and updates for data through the end of August. If there are any other pictures you want, feel free to suggest.
As a closing remark, I'm really happy and intrigued by these pictures. I didn't know what they would look like when I started... but they do tell a story that makes sense, and also help me think about the season differently.