Thursday, February 10, 2011

Why Base Stealing Might Actually Work

So I read Moneyball by Michael Lewis, and it is a very good book. I would recommend it to pretty much anyone. It was the story about Billy Beane and how he managed the Oakland Athletics to winning seasons without the typical winning salary. It was fun reading about how Beane beat the system during the regular season. But it got me thinking of baseball statistics. I mean this book was right about a lot of stuff. It is very smart not to trade outs for movement along the bases by way of sacrifice bunt and sacrifice fly, and on base percentage says more about a player than average. But I think that in a sense, they were wrong about base stealing. Now base stealing actually goes against the idea of trading outs for movement along the bases, but in some situations it actually may help you score a run with a small chance of getting out. Much of the strategizing for the Oakland A's was based on the statistics. They did not use the traditional eye of a scout to judge talent. Winning was based on having better stats than the opponent. So I will use statistics to see if I can make base stealing relevant for teams like the A's.

At first I tried to use what the book said. You know the probability that a player can steal a base by number of successful steals over number of attempted steals. You can also get a rough sketch of the odds of a player scoring from their given base. Example: a player on first has a .25 chance of scoring while a player on second has a .5 chance of scoring. So based on these made up statistics, you could determine if a player should steal bases on any given attempt. So the player must be able to steal at least 1/2 of the time in order for it to benefit the team. But I ran into the problem of the catcher. If you remember Ivan Rodriguez in his hey-day, he was able to peg a lot of runners. He was a base-stealer's worst enemy. So now I had to consider the catcher. If he pegs 50% of base-stealers, is it still worth it? I started considering different types of base-runners and catchers as either good or bad, but then it became very biased and judgmental. How can you determine how good someone is without considering their traits? Since appearances may lie, I had to rely on the numbers.

I finally realized what to do. What matters is not the success of the stealer or thrower. It is the time it takes for both the ball and the runner to reach second base. So what I thought about doing is normalizing the times for both players. Since there was a lot of data, you could use the central limit theorem to find the mean and standard deviation. This may seem like a daunting task, but there are programs (hint hint excel) that can do these equations very quickly. So then you could find where the two normal curves overlap. And instead of using the scoring probabilities, you could use on base percentage. The probability that you get to that base on the person's swing versus the probability you take it yourself. If the probability of stealing the base is higher than OBP, then you can take it and of course vice versa.

So that's the long and short of it. Really nothing that special, probably won't be noticed by any major league teams. But still voicing my opinions one paragraph/sentence/word at a time.

No comments:

Post a Comment