<$BlogRSDUrl$>

Wednesday, December 17, 2003

STATISTICAL HAIRBRAINEDNESS 

I know people have probably thought about it before, but here goes...

I've known all along that the good ol' formula for Earned Run Average is
[(ER/IP)*9].

Basically, the formula cranks out what a pitcher would give up if he pitched a complete game. I've sat and thought about this many a time since pitchers barely ever pitch complete games anymore. If a starter goes seven solid innings, it's considered good and even lengthy nowadays. Furthermore, ERA is still more applicable to starting pitchers than it is to relief pitchers.

ERA to me means almost nothing when used with relievers. Do I really want to know what the pitcher is going to give up over nine innings, an amount of innings which may actually amount to anywhere from four to nine outings?

So I thought about it more today and realized it may be easier for me to adapt the ERA formula simply by dropping the multiplier--
[ER/IP]

If anything, this would make it easier for me to discern the likelihood of how often a reliever is going to give up runs.

For instance, let's say Kazu Sasaki has given up 58 earned runs in 30 innings (don't put it past him).
The conventional ERA formula would chug out a number of [(58/30)*9]= 17.40
The simple ER/IP ratio would be (58/30)= 1.93
Assuming Sasaki pitches one inning each outing like he usually does, one could expect him to give up 1.93 earned runs per outing.

Another example --
Eddie Guardado gave up 21 earned runs in 65 1/3 innings in 2003.
ER/IP = 0.32
If he pitches an inning every time out, he'll give up a run an average of every three or four outings.

Another --
Arthur Rhodes gave up 25 earned runs in 54 innings in 2003.
ER/IP = 0.46
If he pitches an inning every time out, he'll give up a run an average of every two or three outings.

And lastly --
Shigetoshi Hasegawa gave up 12 earned runs in 73 innings in 2003.
ER/IP = 0.16
You could take this one of two ways, since Shig was both a setup man that threw over an inning at times as well as a one-inning closer. As a one-inning guy, he'd give up a run every six or seven outings. As a two-inning guy, it'd be every three or four outings. As a combination, well, I don't have time to do that right now.

Interestingly enough, the multiplier on the ERA formula could be used to tabulate ratios for conventional starts from Joel Pineiro or whatnot.
Joel Pineiro gave up 89 earned runs in 211 2/3 innings--
ER/IP = 0.42
...but let's crank this out for, say, a typical seven-inning start.
[(ER/IP)*7] = 2.94
Assuming the seven-inning start, one could say that Joel Pineiro gives up an average of three runs a game. His standard ERA is 3.78. So yeah, you can mess around with the multiplier and get different numbers for different amounts of innings. I'm just saying that none of these guys throws nine innings on a regular basis. So why use nine-inning stats?

So yes, ERA as we know it is convenient and universally accepted, and this ER/IP ratio would give us (or at least me) a better grip on what I think a reliever will do. Of course, I'm sure something will develop out there (if it hasn't already) where huge spreadsheets that show [(ER/IP)*N], with N being the number of innings to be pitched in a game.

I'm sure there are many flaws to the ER/IP ratio, but it's just something I thought I'd throw out there.

The ER/IP ratio may be total hogwash, but my mind just doesn't like to jive with ERA's that are based on nine innings when they're used on relief pitchers that don't work anything close to nine innings at a time. That's my beef, that's all. You don't have to use ER/IP if you don't want to. I just think I might -- nothing more, nothing less.

Thoughts?

/ Click for main page

This page is powered by Blogger. Isn't yours?

Click for Sports and B's 

home page