Wednesday, October 10, 2007

Do errors have park factors?

In my earlier analysis of IBL park factors, you may have noticed an apparent anomaly. Gezer's park factor for hits is 10% higher than Yarkon's; its factor for doubles is 30% higher, for home runs 170% higher, and for total bases 30% higher. But the factor for runs is only 5% higher. Despite being much better for batting, Gezer did not produce that much more runs than the other IBL parks. How is this possible?

Of course, Gezer also had a 14% lower factor for walks than Yarkon, and no triples, but that hardly seems an adequate explanation.

Now that I've tabulated errors, the explanation is clear. Gezer had the lowest error rate of the three parks, with Yarkon the highest:

Using the same technique as before to calculate park factors for errors, we get the following:

Gezer 0.81
Sportek 1.00
Yarkon 1.20

Does this make sense? Is it at all reasonable that ballparks should affect the error rate?

Of course it is.

First, for the most part, errors can occur only on balls in play. With Gezer's higher home run and strikeout rates, it lags the other fields in balls in play by about 5%.

Still the error rate per balls in play ranges from 7.1% at Gezer to 9.3% at Yarkon. (Wow, that's a lot, no?)

Most of that can presumably be attributed to the size of the playing field. In Yarkon's (relatively) large outfield, flubbing the throw from the outfield - or the catch itself - is more likely. In Gezer, there isn't as far to run to go after a fly ball, and there isn't as far to throw to get it to an infielder.

Calculating park effects for errors per ball in play, we get:

Gezer 0.85
Sportek 0.98
Yarkon 1.17

So yes, errors do have park factors. Or perhaps parks have error factors. Either way, in the IBL, it's significant.

No comments: