Quote:
Originally Posted by kq76
Just looking at your xFIPs above, which is probably the best stat of the 3, they're all 0.3 and lower the 3rd time through. A 0.3 drop is nothing to be alarmed about and I'd much rather take that drop than tire out a reliever that I might want to really use tomorrow, but can't because I used them today. A big part of the game, of any game probably, is (or at least should be) about using your best players as much as possible. No sane NBA fan, for example, would, in a non-blowout win, say, "you know, Michael or Steph, they're probably tired by now, we should bench them the last few minutes of the game". That person would be so ridiculed for taking that stance that they'd be hesitant to comment on basketball ever again. But what do some people say in baseball? "Oh, no, take out that borderline hall of famer late, he might be tired. You say he hasn't shown any signs of being tired yet? Doesn't matter. Our statistics tell us to expect a dip in performance the 3rd time through". That's just nuts.
|
It's a drop of more than .3 relative to their early performance though, since the overall includes the "3rd and 4th times through" as well. And again, the overall also includes the days they got shelled while the 3rd/4th presumably only includes their good days.
As well, while I agree with you that normally xFIP is the one I would care most about of those three (which is why I included it), your point in the second half of the comment intrigues me and makes me wonder if that is also flawed. As you point out, it's certainly possible that a major part of the tiredness penalty is giving up much harder contact- which would track with a pattern where all of them see an uptick in HRs per FB (the rough difference between FIP and xFIP), since while in a vacuum xFIP says that HR/FB is mostly noise, if there *is* a loss of 'Movement' (to borrow OOTP vernacular) either when fatigued or when the batter has seen you twice already, then some what xFIP masks in those situations might be signal instead of noise.