THE BOOK--Playing The Percentages In Baseball

Monday, March 15, 2010

Park factors and weather

It is said that one of the reasons for year to year variation in park factors is weather.  For example, one year could be a particularly hot or cold year and thus a park can be more of a hitters of pitchers park than it is over several years.  Same thing with wind.  There can be other things that affect a park’s true factors from year to year besides weather of course.

Because there are much larger random fluctuations from year to year than those caused by weather (and other things), we usually use multi-year park factors when doing some sort of park adjusting.

Out of curiosity, I looked at average yearly game-time temperatures at all the parks from 00-09, at least according to retrosheet, and compared them to each park’s multi-year (00-09) average.

The SD of temperature per year per park is around 1.70, which means we can reasonably expect variations of up to 5 degrees or so.  That is not going to terribly influence a yearly park factor, but it will influence it a little in some cases.

A degree in temperature is worth around .033 runs per game, so if a park is “off” by 3 or 4 degrees in any one year, it will change the run factor by around .012.  IOW, a neutral park will have a 1.012 RF if it is warmer than average by around 3 or 4 degrees.  The home run park factor will presumably be affected much more than that.

Just last year here are some example of parks whose average game-time temperatures were more than a few degrees wamer or colder than their 10-year (00-09) average.

Team Avg. temp. 09 temp.  Effect on park factor
ARI 77.8 83.4 .018
CHC 69 63.9 -.019
STL 77.6 74.4 -.012
BAL 79.2 74 -.018
CWS 69.3 64.5 -.016
TOR 72.3 68.4 -.014

Also, one of the reasons for the low run scoring the last 2 years, besides less PED use (presumably), is that it was a little colder in both the NL and AL in 08 and 09.  Last year, it was .5 degrees colder in the NL and 2.1 degrees colder in the AL, compared to the 00-09 averages.  (So much for global warming.) That alone would reduce run scoring in the NL by .017 runs and in the AL, by .07 runs, not much but something.

For the record, according to my research, the NL and AL combined lost around .11 net runs in talent from 08 to 09.  (Oddly, OPS (SLG only) went up a little from 08 to 09, but run scoring went down a little.)

Anyway, if you work with park factors, I suggest that your adjust each year by temperature.  For example, if your multi-year run factor for Chase Field is 1.08, since last year was a particularly hot year (they might have just kept the roof open more to save money), you would use a run factor of 1.10 for 2009 rather than 1.08.

I might look at the same thing tomorrow for wind.  Unfortunately, the wind data that are recorded for parks are notoriously unreliable, at least as far as the actual wind on the field is concerned.