THE BOOK--Playing The Percentages In Baseball

And here are the overall totals:

RUNS_SCORED RUNNERS_DRIVEN_IN RUNS_PARTICIPATED_IN EVENT_CD
0.273 0.223 0.496 1B
0.426 0.432 0.858 2B
0.602 0.639 1.242 3B
1.000 0.596 1.596 HR
0.001 0.000 0.001 SO
0.012 0.026 0.038 OTHER
0.253 0.021 0.274 BB_HBP

That last column is the reason that R+RBI-HR does a decent job in describing a player’s offensive performance.  It correlates pretty nicely with Linear Weights. 

There are a few other pieces of data that allows you to make some inferences.  For example, 27% of the time you hit a single, the batter eventually scores, but it’s 25% if reaching by walk (or hit batter).  Reason?  Walk are non-random, and are issued disproportionately with two walks, when they do less damage.

Triples score .64 runners, while HR score .60 runners.  Reason?  HR are non-random, and are issued disproportionately with bases empty, when they do less damage.

One thing which should be kept in mind by those who say RP overvalues power and undervalues BB, is that not all RBI are the same.

Here are the approx number of bases the average runner who was driven in advanced on an RBI event (data from 1985 season):

BB or FO, 1.07
GO, 1.20
1B, 1.56
2B, 2.07
3B, 2.25
HR, 2.35

David, perhaps I should be clearer.  RPI undervalues walks with respect to the batter’s contributions in generating a run.  And similarly, RPI overvalues HR with respect to the batter’s contributions in generating a run.

If you want to argue that you want to give the HR more weight based on the number of bases he moves the runners over, that’s fine.  That’ll tell you exactly what it is that you are asking it to tell you.

It will not tell you any better about his contributions to generating the run.

Sure Tango, my post was poorly worded. I simply was saying that, if you look at your RP values for the HR and BB, it is as you say. But if you additionally adjust those values for the greater value of an RBI on a HR compared to an RBI on a BB, the ratio will get even more lopsided, and properly so.

Which means that if you want RC, then use RC. If you want to use R and RBI, then don’t try to make it into RC with an awkward series of adjustments.

So, I think we’re in agreement.

I updated the main file to include whether it was a DH park or not.

I also put the totals overall here:
http://tangotiger.net/retrosheet/reports/runs_rbi_by_lineup_tot.html

Tango,

Is OTHER just what didn’t fit in the rest of the categories (outs, etc.)?

Right.

I’m surprised that BBs (.25) lead to a run nearly as often as 1Bs (.27).  Last year, 38% of BBs came with 2 outs, vs. 30% of singles; 28% of BBs came with zero outs, vs. 36% of singles.  I would have expected that to result in a lot more runs scored after a single.  Is there an offsetting factor?

No, that sounds about right.  Use the following as the chance of scoring by outs: 39% (0 outs), 26% (1 out), 13% (2 outs), and see if your numbers match.

Yup, that works.