Tuesday, August 04, 2015
Selection bias strikes again
?In Pizza's article, MGL has the first comment, and focus on his last sentence.
I agree with MGL on the selection bias. That bias is huge.
Suppose you have a reliever that enters the 7th inning, expecting to pitch 3 innings (face 10-15 batters). If this reliever gets rocked in his first inning, facing 8 batters, there's a strong change that he won't make it to the 8th inning to face his 12th batter and remain in the pool. He's now been removed from the sample.
So, we should expect that any long-outing appearance will not have any disaster games in there, or at least a disproportionately lower one than expected otherwise.
As for short-outings: if you look at ONLY 3-batter outings, you will find something remarkable, and that is a FANTASTIC outing. Why is that? Because if a pitcher is set to pitch only one inning (as is most often the case) and if he faces 3 batters, then guess what: 0 runs, 0 hits, 0 walks (pretty much). So, focusing on short-outings where the pitcher faced at most 6 batters will have a lot of these perfect innings, that you otherwise wouldn't get.
Basically, you have to be REALLY careful with selection bias here, especially since you are choosing your pitchers AFTER THE GAME. Any time you do that, you have to be extremely careful. And even then, you really have to address the issue of bias.
Recent comments
Older comments
Page 1 of 151 pages 1 2 3 > Last ›Complete Archive – By Category
Complete Archive – By Date
FORUM TOPICS
Jul 12 15:22 MarcelsApr 16 14:31 Pitch Count Estimators
Mar 12 16:30 Appendix to THE BOOK - THE GORY DETAILS
Jan 29 09:41 NFL Overtime Idea
Jan 22 14:48 Weighting Years for NFL Player Projections
Jan 21 09:18 positional runs in pythagenpat
Oct 20 15:57 DRS: FG vs. BB-Ref
Apr 12 09:43 What if baseball was like survivor? You are eliminated ...
Nov 24 09:57 Win Attribution to offense, pitching, and fielding at the game level (prototype method)
Jul 13 10:20 How to watch great past games without spoilers