THE BOOK--Playing The Percentages In Baseball

Still surprised Fangraphs hasn’t done park-adjusted wOBA or all-out wOBA+...

I am not in favor of wOBA+.  I am in favor of wRC+.

Remember that you can get from wOBA to wRC very easily.  So, to get to a “+” stat, you really should go with the stat that makes the most sense.  And in that regard, it’s wRC.  The scale is appropriate for wRC, not wOBA.

With park adjusted wRAA as a component of WAR, we already offer the practical equivalent of park adjusted wOBA - it’s just expressed as a counting stat rather than a rate.

wRAA is park-adjusted?  I didn’t remember that.  Well, then, it’s a snap to convert to wRC+:

a = league runs per PA
b = wRAA/PA
c = a/b + 1
d = 100*c

So, suppose someone is +30 (adjusted) runs, with 600 PA.  And the league average is .12 runs scored per PA (say 4.56 runs on 38 PA).

a = .12
b = 30/600 = .05
c = .05/.12 + 1 = 1.417
d = 142

ta-da!  easy right?

Sean and David are both pretty responsive.  So, if you want Fangraphs to implement this, then tell David:
http://www.fangraphs.com/blogs/index.php/contact/

I think it would be a good addition.

AAAAAAANDDDDDDDDDD, if Sean also implements OPS+ the way I describe it, then you will find that Fangraphs’ wRC+ and B-r.com’s OPS+ will be on similar enough scales that they could be somewhat interchangeble.

A 150 wRC+ for Junior on Fangraphs would be a 150 OPS+ at b-r.com .

It’s up to you readers.  Email them, and tell them you want this done.

The “Batting” component of WAR is the park adjusted wRAA, wRAA itself isn’t actually park adjusted.

I think this should be pretty easy to do, at least in the player pages.  Sometimes it’s tough to get park adjusted stats into the leaderboards, but maybe I’ll go ahead and fix that.

-David

I’m also wondering if to just make things less confusing we could call wRC+, wOBA+.  As in your calculation above is the “correct and official way to calculate wOBA+”  The real difference in wOBA and wRAA/wRC is that the latter includes a player weighted IBB and of course the scale right?

No the difference is that one will return a value of 150 and the other is 125.

Take for example, Derek Jeter.  His wRC is 122.4, wRAA is 36.6 on 716 PA and his wOBA is .390.

So, the average RC for the league would be 122.4 minus 36.6 = 85.8.  That puts Jeter’s wRC+ as 122.4/85.8 x 100 = 143

His wOBA is .390 divided by I guess a league average of around .330 = 118

See?  The scale is different.  That’s why Sean does OBP/lgOBP + SLG/lgSLG - 1.  He gets the scale to double.  That is, if the OBP/lgOBP is 118 and the SLG/ljSLG is 118, then his OPS+ will be 136.

And that’s close to the 143 that we “really” want to see.  OPS+, as Sean calculates, really is on a scale for RC+.

Park adjusting wOBA isn’t that difficult. There probably is a simpler method than the one I devised. First, you park adjust the RAA:

PADJ RAA = ((RAA + ((Lg Runs/Lg PA) * PA)) / PADJ) - ((Lg Runs/Lg PA) * PA)

where PADJ is the park adjustment, and

PA = AB+BB+HBP+SH+SF

Once you have the park adjusted RAA, you can park adjust wOBA:

PADJ wOBA = (((PADJ RAA - RAA) * wOBA Multiplier) + (wOBA * PA)) / PA

where PA = AB+BB+HBP+SH+SF

Why not just do it additive?  If a park adds 30 runs per 6000 PA, then subtract .005 runs per PA.  Didn’t we have this discussion last year?

Yes we did have this discussion last year about additive park factors. While an additive park factor may work better for park adjusting RAA, I don’t think it is quite enough for park adjusting RC. I think a multiplicative park factor is the better choice for park adjusting runs created.

Tango, I am confused by your example in #4. I might be splitting hairs, but isn’t your example wRAA+, not wRC+.

terps/10: I don’t see how it’s a “better” choice necessarily.  I think Brian showed pretty well that additive was, if not better, certainly not worse.  Seeing how the additive is so much easier to calculate, why not stick with the easier way?

***

You cannot have wRAA+.  How could you?  The league wRAA is 0.  In #4, I’m turning wRAA into wRC, and then dividing by the lg wRC.

I’ll refrain from starting another debate about additive and multiplicative park factors. I’m not sure which method is better. Patriot showed me last year that outs (or games) is the proper denominator for multiplicative park factors based on runs. I would think that this also applies to additive park factors.

I should of realized that wRAA+ would be impossible. You can’t divide by zero. I guess it was this part of your example that led to my stupidity: “suppose someone is +30 (adjusted) runs, with 600 PA.” I interpreted this as runs above average. Anyway, it is a very unique idea you have here Tango. I like it.

OK, you did mean RAA by this statement: “suppose soemone is +30 (adjusted runs, with 600 PA).” You added the LW/PA to the lgR/lgPA and then divided by the lgR/lgPA. I’ll shut up now so you don’t think I’m an even bigger idiot.

I think in #4 it’s supposed to read c = b/a + 1, which is consistent with how the sample calculation is done later in #4.

Good eye!