from Amazon
Tuesday, August 22, 2006
Jarvis + Retrosheet = Useful Data
John Jarvis has expanded his team summary pages going back to 1957:
http://knology.net/~johnfjarvis/stats.html
Just to give you a thought…
He compiles baserunner advancement (how often do you go from 1b to 3b on a single), for every team and every year. One of the things that a Markov chain depends on is the state-transition probabilities (the other is the event frequencies, which is typically known).
So, I would always assume the same state-transition probabilities, regardless of park or run environment. This clearly is not correct, since a team in a low run environment would certainly try to go for the extra base more often (and a park with a high run environment should force the runners to be more conservative).
How true is this, and how much of an impact is this? Well, with a little work, we can find out.
And, the 1957-2005 time period covers very low run envioronments (the 60s) and very high ones (the 90s). It covers pitchers’ parks (Astrodome) and hitters’ parks (Coors). Jarvis’ site gives us lots of data to analyze.
The one thing I wish he would do, and maybe I should just tell him, is that for every offensive category, he should have an opponents category. And, as a wish list, a breakdown of home and away.
We need more John Jarvis’ out there to publish data like this. He can’t do it alone.