In thinking about Michael Young's chances of continuing to post 200 hits a year, it struck me that one thing the various projection system don't tell me is the probability of the various stats they are predicting. So today I've been working on a probability distribution that should give us a clue to questions like, what is the probability of a batter reaching at least 500 AB in 2007?

The study looks at batters since 1901, excluding pitchers. Two distributions were developed. One asks for the probability of an at bat level given the players at bat in the previous season. The other asks for the the probability of an at bat level given the player's seasonal age. By at bat level, I perform an integer divide by ten on the number of at bats, basically putting 10 different values into the same bin. Bin 44 represents everyone with at bats in a season between 440 and 449.

To get the final probability, the two individual distributions are averaged together, weighted equally. It's a simple way of mixing the probabilities, but the results look pretty good:

The top of the table is loaded with players who bat a lot every year. The bottom of the table has older journeymen and some young, unproven players. It's obviously not perfect. I might get better results if I play with the weights more. I might look at a three year weighted average of at bats. This is pretty simple, however, and I really like the order of the list. When you see a player projected to get at least 500 at bats this season, this gives you a way of validating that number.

I'm going to think about a way to make this part of the Day by Day Database. For example, you can use the data to figure out which players are most likely to gather between 200 and 300 at bats. And this should be appliciable to other stats as well. Stay tuned.