Features:
- Total value estimates (in runs) for every MLB position player, based on hitting, fielding, and their position.
- Fielding measures are based on the average runs saved according to zone rating and revised zone rating.
- Pitchers are listed according to RAR (base runs saved above replacement) and FIP-Runs (a defense-independent pitch stat estimating runs saved above repalcement).
- League differences are taken into account.
- Closers get a leverage-based bonus (though it's admittedly rough).
I love you.
ReplyDeleteAny chance of also including a runs-above-average for offense number other than just RAR? For some uses, it's nice to have fielding, position, and hitting all compared to the same baseline (even if it's not completely technically correct) and then add in replacement-level credit separately.
sky,
ReplyDeleteI think Justin is taken so the 'i love you' is a little strong . (yeah, lots of sarcasm in there in case it wasn't detectable).
@Sky,
ReplyDeleteI love you.
You're not getting my bud lite.
Any chance of also including a runs-above-average for offense number other than just RAR?
Sure, I can probably do that. I know I probably should do it all the time (in addition to RAR) so I'm measureing something that's a bit more real than vs. replacement level. But columns can be hard to come by.
Until I get it going, though, you should be able to reconstruct RAA pretty easily--just use a pro-rated 2.0 or 2.5 wins adjustment for the NL or AL, respectively. -j
I don't want to impose, but would it be any trouble at all to refigure the RAR numbers based on PrOPS rather than OPS? I'd love to get a gauge on how lucky the redlegs have been this season. It seems that everyone's OPS is below expectation.
ReplyDeleteBrad, unfortunately, it's not really possible to do that. RAR is based on linear weights, which are calculated from raw events (i.e. singles, doubles, homers, walks, etc), and not OPS. I can't accurately back-calculate those event frequencies from PrOPS (or OPS, for that mater).
ReplyDeleteWhat we'd need to do is estimate the frequency of each event based on the batted ball data that OPS uses, and then use those estimates to calculate a "predicted" RAR. That should be possible to do, but it would take a fair bit of work. It'd be neat, but I'm not really up to doing it right now. :)
-j
Fair enough. I guess I didn't really look at how RAR is calculated. What kind of sample would you use to calculate weights for singles, doubles... based on batted ball data, i.e. how many seasons, what park adjustments, etc?
ReplyDeleteI think the calculation would be fairly simple once you decided on a sample.
Nice work. Is there any kind of park adjustment being made with these numbers? Thanks!
ReplyDeletevr, Xeifrank
Yes, all runs numbers are park adjusted using Patriot's 5-year regressed park factors. -j
ReplyDeleteJustin, that's great. Could you post a link to Patriot's park factors data, if it's public?
ReplyDeleteThanks.
vr, Xei
Here you go!
ReplyDelete-j