FOR THE LOVE OF DOG: IRS% for relievers

[Righty Groove]

Originally Posted by Fyrestorm3

Technically, yes you can. Sure, it's always a small sample size. Sure, there's a lot of luck involved. Sure, the stat can swing wildly from season to season. But if I've got the bases loaded, 2 outs, with a 1-run lead, I would feel more comfortable bringing in a pitcher who has allowed, say, 4 of 20 IR to score on the season, as opposed to 10 of 20. It's very similar to bringing in someone who has a 65% ground ball ratio when you need a double play - of course there's no guarantee that he's going to give you a grounder, but you'd rather have that guy in than someone with a 40% ground ball ratio.

Either way, though, the fact that it's a luck-based stat is not a valid argument for opposing its inclusion in OOTP.

The fact that he allowed only 4-of-20 IR to score means nothing as a predictive stat. In that situation you described, you should bring a high-K%, low-BB% pitcher, and/or a flyball-heavy pitcher, because these stats are predictive, i.e. mantained from year-to-year, past-can-reasonably-predict-future stats. LOB% and IRS% are more descritive then predictive.

Not a slight on you at all, though. It's just that miss-used statistics really bug me. No statistic is bad, only interpretations are bad.

Either way, I see no reason NOT to inclued IRS% in the game.

[Fyrestorm3]

Originally Posted by Righty Groove

Either way, I see no reason NOT to inclued IRS% in the game.

That's really all I'm trying to say. IRS% may not be predictive, but it's nice to know.

[The Wolf]

Originally Posted by Fyrestorm3

That's really all I'm trying to say. IRS% may not be predictive, but it's nice to know.

And it has some limited use. But mainly it's informative.

[BeancheBlanco]

Originally Posted by The Wolf

It's mostly luck. See Baseball Prospectus | Ahead in the Count: Why SIERA Doesn't Throw BABIP Out with the Bath Water

This "study" is full of pseudomath and terrible assumptions. In short, using their method of calculating luck, every single statistic in baseball would be considered mostly luck (most noticeably batting average since batting average is nearly perfectly correlated with BABIP). I don't know why BABIP gets so much attention. It's really just batting average after you take out the strikeouts and so nearly all conclusions about BABIP can be made about BA.

As for their calculation of luck...you can't simply force the variables into being identically distributed and divide that standard deviation by the actual standard deviation. That's extremely oversimplified. A better way of going about it would be to calculate the variance in BABIP for each specific player over a specific age period like 25-30. Then average the variances of a large sample of players and find the variance in that average. This number will undoubtedly be more accurate and much smaller, as I expected. Also, the way they calculated it was wrong, even if you use their method. They calculated the idd variance using n=610 since there were 610 balls in play in a season on average for a qualifying pitcher. They then compared this to the standard deviation of the entire 6 year period using all of the data form all of the pitchers. This is incorrect. Essentially they are comparing the standard deviation in one season to the standard deviation over a 6 year period, which balloons the "luck" percentage. The n value should not be 610. It should be the total balls in play over the 6 years of data for all the pitchers, which is a way, way, way larger number than 610. This would decrease the "luck" variance considerably.

Also, for their calculation on the effect of defense and stadium, they oversimplified the problem again. They calculate the effect of stadium and defense on BABIP by averaging the BABIP of certains teams over a 7 year period and finding the standard deviation. That doesn't make sense because each year, teams change. The standard deviation of team BABIP will approach 0 as you use data further back in history because defenses fluctuate from good to bad. In theory, all teams will have equal defenses if you look at baseball from its beginnings...i.e. a standard deviation of 0. Does that mean defenses have 0 effect on BABIP? No. It means no team has an inherit advantage over another team in defense, if you look at a 100 year period.

[The Wolf]

/fail

[BeancheBlanco]

Originally Posted by The Wolf

/fail

no offense but I feel like all of your posts are 3 word, condescending remarks. Why don't you actually address what I said? I put time and effort to read your link and make my own conclusion from it and for all I know you didn't even read my post

[texasmame]

Sweet furry kitten balls!

I don't care if it's luck, science or voodoo math - I want to see it reported in the game!!!