Baseball Musings
Baseball Musings
July 31, 2005
Fair Run Average

Fair Run Average gets an article in the New York Times today. Basically, it's a way of correcting ERAs for bad relief pitching.

On average, putting runners on first and second with no outs - the mess Foster made - leads to 0.9 more runs than a typical inning. One of the runners often scores. Both of them usually do not.

In a fairer world, the main pitching statistic would have charged Foster for the damage he did, 0.9 runs, but no more, whether or not the two runners ended up scoring. Since they did score, Kolb would have gotten the rest of blame: 1.1 runs' worth.

Had the runners been stranded, Kolb would instead have received credit for his good work. His line could have been 1 inning with minus 0.9 runs allowed. His runs-allowed total for the season would actually drop.

The problem with calculations like this is they change season-to-season. So FRA is a moving target; as the probability of scoring changes with each game, a pitcher's FRA may change without him pitching. This is one of those statistics (like linear weights) which can only be accurately calculated after the season is over.


Posted by David Pinto at 10:28 AM | Statistics | TrackBack (0)
Comments

When I've been working in this vicinity, I just prorate ERA... since ERA charges the "starter" with the runs which he left on base, not the reliever. I deal with the actual, not potential runs, and charge the starter .75 for a runner on third who winds up scoring, .5 for a runner on second, and .25 for a runner on first, charging the balance to the reliever... of course if one reliever advances somebody and another scores him it has to be split up even further, but you get the drift...

The one really impossible situation under this system is the reliever who comes on with bases loaded, nobody out. and gets a 6-4-3 double play and a strikeout; he's done the best you can legitimately ask of him, yet I still wind up charging him a quarter of a run... Still, its pretty simple to figure out in season anyway, and gives you most of the insight you want...

Posted by: john swinney at July 31, 2005 11:10 AM

I'd think you'd want something linked to outs. E.g. relief pitchers is assigned:

- 0 outs: no runners
- 1 out: runner on 1st only
- 2 out: runners on 1st & 2nd

Relieved pitcher is assigned the remaining runners.

Posted by: Jason at August 1, 2005 10:17 AM

I wouldn't worry too much about getting it perfectly. Whether you give .26 or .28 runs for every runner left at 1b with 1 out, what difference will that make over 35 starts? I'll guess that leaving runners on base will happen in less than half the starts, and you'll be off by .02 runs or so at the most. So, whether you say x number of runs for 15 left-runners, or x+.02 runs for those 15 runners, that difference is still less than half a run.

Btw, this computation being described is basic LWTS by the 24 base-out states, and is nothing new.

Posted by: tangotiger at August 1, 2005 03:39 PM
Post a comment









Remember personal info?