Monday, June 09, 2008

Maybe It's Not the Whole American League...

With the unbalanced schedule, one thing I was going to look into this off-season was calculating some sort of a division factor, as I had noticed a surge in offense in the National League East the past couple of seasons. Just as playing half of your games in Coors Field will inflate numbers, playing almost half of your games against the same small group of teams could have an effect on statistics. If the other teams in your division have great hitters, your pitchers will likely have worse numbers. If every team in your division has great pitching and defense, your hitters will likely struggle. If you have to play more games than any other opponent in Petco Park, that will probably have an effect... except that is offset by the same amount of games played in Coors Field.

I wish I would have gotten around to looking at this before the season, because I could have taken credit for thinking Miguel Cabrera and Edgar Renteria would have worse seasons as they move into a tougher division to hit. While I think that is part of the reason they have failed to live up to expectations, there likely is more to the equation.

I first started to look into this when I saw some fantasy baseball transactions over the past week. Travis Hafner was dropped in one of my leagues; Victor Martinez was benched for Dionar Navarro in another. I had high hopes for Billy Butler in one of my leagues, only to see him tank and end up in AAA. One owner was thrilled to have Delmon Young finally hit a home run this week. Oh yeah, this Tigers offense isn't exactly on pace to put up 1000 runs. Something seems to be amiss in the American League Central.

Interestingly enough, this is not the case. I wrote that whole introduction before crunching the numbers, and the results surprised me.

I'm about to introduce a statistic called Divisional Effect. This will be similar to a team's park effect, where the team's runs scored and allowed within the division are divided by the runs scored and allowed outside of the division, multiplied by 100 and rounded to the nearest integer to make things pretty. The Divisional Effect will have two components: oDE, which measures the effect on runs scored and pDE which measures the effect on run prevention. American League Central teams ended up with a oDE of 105 and a pDE of 100. This suggests that American League Central hitters actually do better within the division. Another look at the depressed numbers of AL Central hitters would seem to suggest otherwise.

I've run out of time for now. I will have more on the Divisional Effect later in the week. I just came up with it this morning and want to refine some things before coming to conclusions.


Lee Panas said...


The divisional effect is a good idea. I'm curious to see what you come up. I just wonder whether one year of data will be a big enough sample size. Park effects require at least three years of data to be useful. One year for the divisional effect might be ok though because there are more games involved.


Eddie B said...

Three years wouldn't work for a division, given that the strength of the teams changes so much in one given year. I should have something by mid next week.