Friday, April 23, 2010

Describing Durham Bulls Team Performance

click on chart to enlarge

This morning I put up a chart that was somewhat legitimized by the Pythagorean idea from Sabermetrics. With this chart I invite you to follow me on a true statistical leap of faith. Leaps sometimes have difficult endings, but maybe not this time. We’ll see how it goes this year.

Step-by-step, so you can watch me approach the cliff, here’s where these curves come from.

I started with the idea of the Weighted On Base Average, or wOBA, a measure that is popular among the sabermetrics crowd. Basically, with the wOBA you take every kind of way a batter can get on base (singles, doubles, triples, home runs, walks, hit by pitch, errors), weight each one, and then divide it by the number of times he comes to the plate. Think batting average, but with a better appreciation of overall contribution made.

So far, so good.

But wOBA is mostly used to measure individual performance. My interest is in getting a handle on team performance. And drawing a picture of that performance.

So, I put together numbers for the Bulls team wOBA. Some sites publish team batting averages and there’s at least one site that publishes team wOBAs, but these are my own computations, so any errors are my fault. Note that one of the reasons the curves wobble around a bit at first is simply that the numbers are very small, the sample size problem is alive and well at this point in the season. But that should not keep me from getting over there close to the edge of the cliff.

Next I built a spreadsheet that recomputed the team wOBA at the end of each game. So game one was .346, then wOBA was recomputed for games 1 and 2 combined and that was .367. Then game 1, 2 and 3 were added together and team wOBA was .347. Thus, each point on the chart for the green line is the team wOBA up through that game. As of April 22, the Bulls team wOBA is .382.

So far I don’t think that I’ve violated any major statistical principle (other than the small sample size problem mentioned earlier). But maybe this does. It seems to me that if team wOBA tells me how the Bulls are doing on the offensive side of things, wouldn’t our opponents’ team wOBA tell us something about how the Bulls are doing on the pitching/defense side? I computed Opponent Team wOBA and those data points are the red line. (Cautionary note: We have only played three other teams so far this year.)

Just as the Bulls team wOBA is changing, and their opponents’ team wOBAs are changing, so is the gap between them. As of today we see a gap of .097 in the Bulls’ favor. It’s been as low as .000 back on April 12.

Is this charting going to be of any use as the season goes on? I’m not sure. All I can say now is that this could be a nifty way to separate out offensive and defensive trends. For example, it looks like our pitching has mostly held its own and our offense has gone nuts over the last several games. Also, as the year goes on we can surely expect less and less variation. Nevertheless, if the gap correlates with other measures (won-loss, games above/below .500, etc) then we might have something really interesting to keep track of over time.

At any rate, if I haven't gone off the edge of a cliff, the view is beautiful from here.

No comments:

Post a Comment