WPA vs. wRAA

by robbbbbb on Jun 16, 2011 8:29 AM PDT

I'm looking at the difference between WPA and wRAA as a tool for measuring leverage.

Background: I've got a coworker who has argued that Ichiro's hits come "when they don't matter." I don't believe this is true and I'd like to test the hypothesis. I think the best way to do this is WPA vs. wRAA. My understanding is that wRAA is a straight linear-weights, context neutral stat that includes plate appearances, stolen bases, and caught stealing. WPA only uses the same events, but is entirely context-dependent. Taking the difference between the two should yield an answer on whether a player's contributions come in low-leverage moments.

I'm using Fangraphs as my data source, because they're awesome.

However, I have a couple of questions on the use of these stats to make sure that I'm doing this properly.

First: Do they, in fact, cover the same events? (Or so close that I can get away with it.)

Second: Which wRAA do I use? There's the wRAA that's listed as such on Fangraphs, but then there's the batting component of WAR listed at the bottom of the page. They're different, and it looks like the batting component of WAR is park-adjusted; should I use that?

Third: What's the conversion factor? I know that +10 wRAA = 1 win. And +0.50 WPA = 1 win. So 10 wRAA = 0.50 WPA, and I can work the conversion from there, yes? But then I start going around about the math and wonder if +10 wRAA gets you from 81-81 to 82-80, and is thus worth +1.00 WPA. (And I'm aware that 10 runs = 1 win is a convenient abstraction, and the actual run value varies a bit from year-to-year. And I'm wondering if that's significant enough that I need to account for it, or if I can use the standard 10 runs = 1 win.)

Or, am I attacking this problem from an entirely wrong angle, and there's a better way to do the analysis? (Or even someone who's done the legwork for me, and I can point to the results?)

Thanks in advance for the help with the questions.

Tweet Comment 7 comments | 0 recs |

Do you like this story?

Comments

Clutch

Under the “Win Probability” section, fangraphs has a column called “Clutch,” which is described as “How much better or worse a player does in high leverage situations than he would have done in a context neutral environment.”

Baseball-Reference.com says Clutch is “The difference between context dependent WPA and the context-neutral WPA.” Clutch = WPA (overall)/aLI – WPA/LI

For his career fangraphs has Ichiro as +6.5 (bref says +5.8). With positive clutch values every year except one (2010).

Furthermore, for the years 2001-2011, Ichiro is number 1 on the fangraphs leaderboard for accumulated Clutch, and #12 all-time (well, since 1974, anyway).

That said, none of this really means a whole awful lot, as clutch skill is thought by most sabre-types to be a (probably) real, but very weak and noisy phenomenon.

by MangoLiger on Jun 16, 2011 9:03 AM PDT reply actions

FanGraphs is great, but I think Baseball-Reference can help you easily win the argument

Just look at Ichiro’s leverage stats. The far column—tOPS—shows that he hits slightly better in high leverage situations and worse in low leverage situations. It’s enough to at least refute your co-workers argument, if not prove that Ichiro is clutch.
Scroll up to B-R’s clutch stats and see that while he’s a tiny bit below average in “late and close” situations, he’s great with 2 outs and runners in scoring position (though that’s probably inflated by the fact that he’s intentionally walked almost one out of every seven times he’s in such a situation.) Otherwise, his tOPS stats are strikingly consistent. Consistency is one of the amazing things about Ichiro.

by yuniform on Jun 16, 2011 9:59 AM PDT reply actions

Just send him this

Any excuse to post it. And see if he still wants to argue.

by Drew_D on Jun 16, 2011 12:25 PM PDT reply actions 3 recs

This is incredible

by Bearskin Rugburn on Jun 16, 2011 1:20 PM PDT up reply actions

The best thing about this is Ichiro's homerun destroys LA and Orange County, not Seattle

clutch

by Bearskin Rugburn on Jun 16, 2011 1:22 PM PDT up reply actions 1 recs

I think before you get into any kind of discussion with your coworker you need to determine how dogmatic they are about this

In other words, if you present a cogent and convincing argument, are they willing to say, okay I was wrong, or will they find some way to continue believing what they always believed? If it’s the latter, and I guess it is, you are wasting your breath.

by Bearskin Rugburn on Jun 16, 2011 1:25 PM PDT reply actions