from Amazon
Wednesday, April 02, 2008
Getting to Know wOBA
With a hat tip to David at Fangraphs for the idea…
What the f is wOBA anyway? If a guy’s walks and extra base hits, relative to each other, are similar to the league average, then a player’s wOBA is simply his OBP. Let’s look at some cases:
I’m taking the 2008 Marcels, if you want to play at home. I don’t have the reached on error numbers, and I’m including IBB for ease. So, this is what the league average is:
OBP: .338
SLG: .426
wOBA: .335
I’m going to add .0025 to all the wOBA for each player, to get them aligned for my purposes here.
There are 0.847 extrabase hit per walks+hitbatters. If I select all players who have between .837 and .857 extrabase hits per walk+hitbatters, I get this:
106 players, .338 OBP, .424 SLG, .337 wOBA
Now, what if add an additional filter to the above list: wOBA of at least .340. In this case, I get 23 players, .355 OBP, .450 SLG, .355 wOBA, with XBH to walk+hitter ratio of .848.
You see, as long as the “profile” of the player remains fairly constant, his wOBA will match his OBP.
Here’s 10 of those 23 players:
PA wOBA mOBP mSLG xbh_to_walk nameLast nameFirst
552 0.364 0.361 0.464 0.85 Tulowitzki Troy
548 0.406 0.396 0.544 0.85 Jones Chipper
539 0.348 0.357 0.425 0.85 Renteria Edgar
500 0.359 0.369 0.437 0.85 Pedroia Dustin
488 0.348 0.342 0.449 0.85 Jenkins Geoff
477 0.364 0.364 0.462 0.85 Hermida Jeremy
463 0.357 0.359 0.451 0.84 Kotchman Casey
438 0.348 0.353 0.433 0.84 Catalanotto Frank
383 0.348 0.341 0.453 0.85 Betemit Wilson
369 0.349 0.348 0.448 0.85 Spilborghs Ryan
You will note that in the case of Chipper, his wOBA is higher than his OBP. In this case, our XBH per walk selection criteria didn’t do a good enough job in terms of getting an average “profile”, since Chipper has a disproportionate number of HR.
Now, what if we select guys with lots of XBH per walk? If we set the selection criteria as at least 1.2 XBH per walk, and at least a .340 wOBA, we get 17 players: .347 OBP, .491 SLG, .363 wOBA, 1.34 XBH per walk. See, in this case, the OBP is not a good way to measure these guys, since they have alot of value in their SLG that OBP simply doesn’t reflect. wOBA however captures this.
If I had selected guys with a wOBA and OBP of between .358 and .368, I get this: 22 players, .363 OBP, .464 SLG, .363 wOBA.
These last two groups are similar. A .363 OBP with a .464 SLG is the exact same thing as .347 OBP and .491 SLG. They are both .363 wOBA. The 16 points of OBP on one side is balanced by the 27 points of SLG on the other side (i.e., a 1.7 to 1 tradeoff).
The bottom line is this: if you see a guy with a .360 OBP and a .360 wOBA, then you know he’s got a “normal” profile of extrabase hits and walks. If you see a guy with a .340 OBP and .360 wOBA, then you know there’s alot of power that his OBP is not capturing. If you see a guy with a .380 OBP and .360 wOBA, then you know he’s not a power hitter. That’s why I like wOBA.
On top of which, by casting it as a rate-like stat, the binomial theory is opened up for us. While normally, you’d figure SD = SQRT(OBP*(1-OBP)/PA), in the case of wOBA, you change the “1’ to “1.1”. It’s a good shorthand, when you work with groups of players of “average profiles”.