Fire Gardy

Last night we could all enjoy a comfortable game where the Twins coasted to victory over the White Sox. The game was actually close coming into the seventh inning, when the Twins batted around and plated 7 runs.

In my ongoing effort to introduce everyone to the WPA graphs from FanGraphs, here’s what it looks like when a team scores a bunch of runs and turns a tight matchup into a comfortable win:

Twins at White Sox, 4/10/2009

As you can see, the Twins and Sox were trading right around 50% through the first half of the game, until Morneau started the 7th with a home run and after that the graph just went to zero for the rest of the game.

Last night I received a question about the second graph below the WPA chart. My first thought was that it represented the WPA change for each play, but it didn’t take long to realize why that doesn’t make sense, given that it’s basically a repeat of the data in the top graph. But, obviously, the label underneath it calls it the “Leverage Index,” which is a term I’m familiar with; basically, it represents how important a situation is to a game, before the outcome (WPA represents what happened after the outcome).

The tallest bar isn’t a perfect representation of how the WPA graph isn’t necessarily tied to the LI graph below it. That bar represents the situation when Josh Fields came to the plate in the bottom of the second, with the bases loaded and two out. The LI bar represents that that’s a critical situation. The spike in the WPA graph indicates that he came through. If he hadn’t come through, the WPA graph would have gone slightly down rather than way up (because it represents post-outcome probability), but the LI graph would still have a tall bar (because it represents pre-outcome leverage).

So today we got to look at another shape of a WPA graph, and learned a little more about the information the graph can tell us. I’ll try to keep posting WPA graphs on my game recaps, perhaps only when they’re interesting or novel.