from Amazon
Tuesday, May 04, 2010
Observational data
Ben/ESPN:
Here’s an example from April 10 of this season. With one out in bottom of the ninth, two-time Fielding Bible Award winner Franklin Gutierrez chased an Elvis Andrus fly ball back into the deepest crevice of right-center field and made a leaping catch to rob a game-tying home run.
...
The Plus/Minus system looks at the ball’s location, velocity, and trajectory to determine the difficulty of the play. Of hard fliners (we group both fliner types together in the Plus/Minus system) to that location on the field in the past year, only 17% were caught. Confirming what our eyes told us, most fielders wouldn’t have made that play. We give him 1 - 0.17 = 0.83 plus/minus points, or “plays”.
Prima facie, 17% seems way too high for that play. Let’s say you have five types of CF:
- 10% great ones (like Gutierrez)
- 20% pretty good ones
- 40% average ones
- 20% below-average
- 10% bad
How often is that play turned into an out? I’m going to guess that if Gutierrez sees that play, he makes it 50/50. The pretty good ones might make it 25% of the time. The average ones 10%, the below-average 5%, and the bad ones never make it. That gives me a weighted average of… 15%. So, yeah, I guess 17% is actually pretty reasonable, given this simple model to test it against.
Ben also says:
Additionally, all of the hard fliners hit to that location (at least those that stayed in the park) went for doubles when they weren’t caught, so we multiply his score by two (0.83 * 2 = 1.66). This is what we call “Enhanced Plus/Minus”, which we use to account for the play’s extra-base impact.
In The Fielding Bible – Volume II, we converted the Enhanced Plus/Minus scores to “Runs Saved”. For center fielders, every 1 Enhanced Plus/Minus point (or “base”, if you prefer to think of Plus/Minus in terms of “bases saved”) translates to 0.56 Runs Saved. Gutierrez’s catch results in 0.93 Runs Saved (1.66 * 0.56).
It looks like a convoluted way to get to where we want to get to. A double is +.77 runs or so, and a double where that double went would probably be more like +.85 runs. But an out would be -.30 runs. So the difference is 1.15 runs. I remember having a problem with how the Enhanced version was calculated. The Linear Weights path gives you a straightforward path there. This doubling here and multiplying by .56 there is… not straightforward. I’m sure Ben can justify the longer more complicated practice. But I can justify BETTER an EASIER practice. It’s just unnecessary to take extra steps when you don’t have to.
Recent comments
Older comments
Page 1 of 344 pages 1 2 3 > Last »Complete Archive – By Category
Complete Archive – By Date