Interesting. From a Moneyball perspective, I think the response to this would be that you need to look at the strategies themselves, rather than the success rate. For example, most of the conclusions are based on teams that swing for contact and hit safely often... but we know (or believe) from Moneyball that reaching base on hits is pretty random. So it's sort of Madden-esque to say that teams that hit safely often will end up scoring more runs than average.

To me, it's much like saying that successful pitchers who try to get batters to hit it to fielders will do better than the rest. Of course this is true in a sense: when batters hit balls right at fielders, the pitcher does tend to give up less runs. But we know (or, again, believe) that pitcher's can't actually control where the hits go.

So we're seeing a difference between batters who swing for contact and hit safely versus those who swing for contact and don't hit safely. The difference between these two groups is the RESULT of the contact - which is random. So it sort of strikes me that the variation among these contact teams isn't due to certain teams being good at hitting for contact, but rather just being fortunate over a given season.

What the heck is this FSC thingy? I'm still looking at this, but I totally agree with you, he's not measuring what he thinks he's measuring. Expanded, if I understand him correctly, the equasion for FSC is

FSC = H/AB * (TPA-BB-SO)/TPA

TPA itself has an AB term in it... this leads to an AB^2 term on the bottom for no particular reason. Doesn't he actually want to measure

FSC = H/TPA

??

Rob - actually, for FSC, I used team batting average times the percentage of plate appearances for a team in which they did not walk or strikeout.

So, in long form, that would be:

[(H/AB)*((AB+BB+HBP+SH+SF)-SO-BB)/(AB+BB+HBP+SH+SF))]

Does that end up with an AB^2 in the bottom?

Yes. Yes, it does. Let's use a slightly different notation here to clarify:

H * ((AB+BB+HBP+SH+SF)-SO-BB)
--------------------------------------------
AB * (AB+BB+HBP+SH+SF)

Okay, taking the first derivative, you end up with only one AB on the bottom, so taking this out to infinity, I suppose you could argue it's close to the same thing. But I guess my question still is -- why do you care to do the whole right hand side of this as opposed to H/TPA? Isn't that the definition of "hitting safely", anyway? Or do you consider (I guess you must!) that sac hits (bunts?) and sac flies also count as "successful"? Even then, assuming you do, why not simplify the whole thing to

(H+SH+SB)/TPA

?? I mean, look at it this way: BB doesn't occur in the denominator of batting average, but it does for OBP. Make it consistent for everybody and use TPA as your denominator, then go back and do your calculations.

Additional note: taking the redefined version of FSC, there's still only a .265 correlation between FSC and runs scored from 1994 to the present day; using the original definition you provided earlier, r= -.014. There is, however, a .575 correlation over that same period for OBP. If there's a correlation between runs scored and FSC, I'm not seeing it.

Got to admit that this makes no sense to me. FSC and FBB are relatively arbitrary measures. There's nothing to suggest they're compararable, particularly in the way they're being compared in this article, or that they're even useful in the first place.

Mike: I have to admit to not reading Moneyball (Gasp, I know, I need to), but does it actually proclaim that hitting a ball safely is random? I've seen the studies for pitching which suggest that (although, on the further studies, I would go with "Doesn't influence as much as previously thought" rather then random).

However, it seems that people who hit safely on contact does not seem to be random. Players with hugh averages generally continue to have high averages, and vice-versa (and also tend to generally have the same strike-out rates). Otherwise, say, all lower-strike-out rate players would have random averages (so players such as, say, Ichiro Suzuki's average would change randomly, as would Rey Ordonez's, wouldn't it?)

Ray, I agree that the formula seems "off" a bit on what he's trying to accomplish. However, I think he wants to keep "BB" out of the final denominator (so anything\TPA wouldn't work). From what I can sense, he's trying to make a percentage out of high batting average teams compared to their plate apperances that result in contact (and make a formula out of that), thus anything divided by TPA would include K's and BB's.

Understand that this is not aimed at you, Steve, but at Lombardi.

Why on earth would you want to remove BB from the denominator?

Multiplying these two things together just looks to me to be completely wrong. Apples to apples, or go home.