Bill James and Accuracy
Note: I inadvertently scheduled two articles on the same day. Rather than reschedule every single blog post, I’m releasing this as a second article today. Enjoy!
I’m up to The 1979 Bill James Abstract in my review. As usual, I’m only reviewing blog posts written about this volume by Rich Lederer, since the originals are almost impossible to find. Of course, if you happen to have physical copies of the original 1977-1981 Abstracts (or the reprints) that you are willing to part with, please contact me.
This section stood out to me:
Now, you know as well as I do that you can’t extract “general truths” from stats that are technically incorrect. And we kind of see that in the case of Richie Hebner in 1978:
There’s no question that he was a better hitter in Philadelphia than elsewhere. In fact, it looks almost like two completely different players.
But, interestingly enough, you don’t see the same pattern in 1977 — despite the fact that he was with the same team (the Phillies) that year:
He hit for better power at home, sure — but the home field advantage he enjoyed in 1977 wasn’t quite as obvious or prominent as it was in 1978.
Go back to 1976, when he was with the Pirates in Three Rivers Stadium (which was very similar to Veteran’s Stadium), and the difference completely disappears:
The Mets picked him up in 1979, and this is how his home and away splits looked:
Maybe the Mets would have refrained from going after him if they realized that Veteran’s Stadium was helping him so much. I think that’s what James was getting after — though it’s hard to tell without having the original text in front of me.
However, that’s really not the lesson to learn here. Hebner’s 1978 home batting record was an obvious outlier. There was no reason to expect him to continue producing at that offensive level.
Hebner was 30 years old in 1978, right in the middle of his prime. He played through 1985, ending his career with the Cubs in a part time role. The Mets probably should have thought twice before going after a 31 year old in 1979 who was just coming off one of the best seasons of his career.
In other words — while park effects do matter, we need to be careful. Sometimes what we think of as a statistical signal is actually just noise.
This matters, by the way. When you’re making a baseball sim, or when you’re going through your own replay project, you need to keep in mind that a lot of the bells and whistles you might be tempted to add in are likely just small pieces of statistical noise.
But, of course, James wasn’t going to attract a major publisher to professionalize his self created book if he started talking about statistical noise.