I think it's worth penalizing for GIDP as well, given they're single-handedly producing the same result as a CS plus a regular out. Depends on whether you're measuring base contributions, or hitting + baserunning. If the latter, reaching on error probably should be included, too... Faster runners cause hurried throws, and all.

The precision police would probably suggest CS should be indexed based on which base was being attempted. Caught stealing 3rd erases two bases, whereas caught stealing 2nd erases just one. The bulk of CS is 2nd base, so this should be fine.

So, uh, someone figured out OPS, and was really bothered by the double-counting of batting average?

This is not a new idea, obviously, and it's not as accurate as some other metrics, but it's not bad for a garbage stat. If a kid or a new baseball fan came up with it, I would applaud.

Holy Total Average!

I came up with something similar to this (though simpler) eight years ago or so, when I was just starting to learn about baseball beyond what you see on telecasts. I didn't consider steals (nor GIDP), though. It was just (TB + W)/PA, which I called Total Base Average.

This doesn't hold a candle to the Snow Index.

Holy Total Average!

The thing that seems to separate this from the annual rederivation of Total Average is the fact that the author isn't presenting it as something new and revolutionary; he's just messing around and having a little fun, which is all that most of us are doing with this stuff too.

Just what we needed, another garbage stat. Hey! Work on pitching and defensive stats instead!

This doesn't hold a candle to the Snow Index

It certainly doesn't licky boom boom down.

Interesting that the creator wanted to ignore sacrifice flies (but still penalize RBI groundouts).

It certainly doesn't licky boom boom down.

####. This may be the best Primer post in 10 years.

####. This may be the best Primer post in 10 years.

seconded.


as for the stat, personally, i don't think it's junk. i think that if everyone grew up with this as the predominant stat instead of OPS, we'd probably be noticeably better off. OPS double-counts batting average, yes, but it also combines two numbers with different denominators as if there's nothing wrong with doing it.

OPS is pretty ingrained in our collective psyches, but if both stats were on the main page at B-ref, i'd probably reference this one more often than OPS.

#12 If I was the stats man at B-ref I'd test whether Total Average modeled run scoring well enough to merit inclusion. It's an awful lot more work than OPS for a truly trivial gain in accuracy (as measured either by correlation with team runs scored or by RMSE). We're talking a correlation of .926 for Total Average versus .922 for OPS. And the difference in the standard error is .21 runs. (Numbers from an old Clay Davenport study. Different time frames give slight different results but the overall conclusion is basically the same.)

It's not as good as runs created (which is basically the same amount of work) for the simple reason that it over-weights stolen bases.

OPS is "good enough" for in season (particularly since it's not hard to make mental adjustments for all of the know weaknesses in the method) and has the virtue of being easy to calculate from generally available numbers.

just for shits and giggles:

by this measure ((TB + BB + HBP + SB - GIDP - CS) / (PA)), jack morris rates at .426, bert blylevin rates at .399, dave stieb at .395, dennis martinez at .424, brett saberhagen at .394, orel hershiser at .395, curt schilling at .404, and roger clemens at .389.