Bernanke on ERA

[1e Corbeau]

Just heard this on NPR; it concerns inherited runners and their uneven effect on ERA.

 

Ben Bernanke, President Bush's choice to replace Federal Reserve Board Chairman Alan Greenspan, is a student of financial as well as baseball statistics. In particular, Bernanke thinks the formula for evaluating the performance of baseball pitchers -- the calculation of the Earned Run Average -- could use some revision.

 

Dwight Jaffee, the Willis Booth Professor of Banking Finance and Real Estate at the Haas School of Business at the University of California at Berkeley, helped Bernanke come up with this idea. He talks about the Fed chair nominee and his ideas.

 

 

Link with audio

[1992casey]

It's an interesting little blurb.

 

As you mention, he wants to see ERA changed somehow to consider the effect of inherited runners.

 

I figure that ERA itself is a long established stat which carries some flaws; this flaw is among them. But due to its long existence, rather than changing ERA itself, I'd simply try to come up with alternative measures and call them something else. That's already going on, but there's always room for more work.

[rluzinski]

Yeh, they will never change the definition of ERA and there's already alot of different measures taking inherited runners into consideration.

[dadofandrew]

I guess I just don't get this. The reliever should be "blamed" somehow if runners already on base scores? So if 2 pitchers come in and get 3 outs, yet one of them comes in with 2 guys on and they score, he did a poorer job?

 

Inherited runers is as flawed a stat as exists, as it blames a pitcher for his bad luck. A runner on 3B/0 outs counts the same as a runner on 1B/2 outs. Also, you find consistent performance in this is rare, much like "clutch hitting". I would say it is mostly good fortune.

 

A good stat might be to take runs scored and divide it by "expected runs" So if a reliever comes into 10 games in situations where the opponent's "expected runs" is 13.5, and allows 8, his number would be 8/13.5, or .593...anything under 1 would be better than expected, I guess.

 

Heck, maybe it's done, and just not getting any publicity.

 

I

[1992casey]

I'd have to look for it, but about a year ago I saw some work on inherited runners that appeared to be a definite step in the right direction as it was centered on base-out states¹.

 

The problem was that the work was nowhere near ready for prime time. While it appeared to be quite valid (accurately measuring what it was supposed to measure), the explanation pretty much required a college math degree to comprehend and the resulting number made no sense in isolation. I felt it was way too difficult for even the sabermetric crowd, much less the mainstream.

 

Then again, the math involved in the NFL quarterback ranking formula is difficult and the resulting number makes little sense in isolation, but fans have bought into it and gotten used to it.

 

---

1. The 24 base-out states refer to the number runners, which base(s) they're on, and the number of outs: Bases loaded, nobody out, runner on first, two out, etc.

[rluzinski]

Baseballprsopectus does just that (I doubt they were the first).

 

LINK

[1e Corbeau]

The reliever should be "blamed" somehow if runners already on base scores?

 

I think the opposite view of the reliever being rewarded for, say, getting two outs while facing only one batter (for instance, a runner on first with one out) is just as valid. While this reliever has at least some influence in inducing a ball put in play (a ground ball? a line-drive?), the removed pitcher is not rewarded with 1/3 of an inning pitched. However, the removed pitcher is also not "punished" with an earned run after he has left the game.

 

Furthermore, what if the man on first is Mike Piazza, and he was intentionally walked?

 

I guess this is why statistics can be fun, difficult, and controversial. And why complexity and accuracy are often directly proportional.

[backup catchers]

I couldn't get the audio of 1e's post but I think I get the point. Looked at the link Russ posted and got a flashback to college stats.http://forum.brewerfan.net/images/smilies/sick.gif I've always felt era for relievers is the equivelnat to batting average for a utility guy. Not a big enough a sample size to determine real value. I would like to see an inherited runner stat totally independant of era. It could an expected run vs. actual runs correlation or something. Probably is something like that already but it just doesn't seem to be something generally used in evaluting relievers. Not by many fans anyway.

[rluzinski]

Here's a Dan Fox blog entry summarizing different ways to "fix" ERA:

 

A More Accurate ERA

[mothershipconnection]

The reliever should be "blamed" somehow if runners already on base scores? So if 2 pitchers come in and get 3 outs, yet one of them comes in with 2 guys on and they score, he did a poorer job?

Yeah, I think it's kinda bogus. If Wise comes in with no outs and nobody on and gives up a single before striking out the side, then Turnbow comes in the next inning with nobody out and the bases loaded, giving up one single before striking out the side, who did a better job? Though Turnbow allowed more runners to score, he did the exact same thing as Wise in a higher-pressure situation, so I can't see any way he's viewed as having pitched worse.