Are the Red Sox Really too Lefthanded?

Which, to me, at least, is proof that he’s not the same hitter. I set the cutoff there because I find far more similarities between years ‘06/’07 and ‘09/’10 than I do between ‘04/’05 and ‘06/’07. I may make 20% of my sample his good year against lefties, ‘07, but you made 27% of your sample his injury year in 2008 when saying "he’s the same hitter from ‘04 straight through to ’10," so I’m not sure I see your issue there…

If CC was truly improving, you’d expect to see better numbers; but we don’t. The past three seasons, Crawford has posted wOBAs of .289, .313, and .306 against lefties.

Again, here’s where we seem to reach a disconnect. To even get to the point where the splits would have any reliability, Carl Crawford would need around 200% more plate appearances! Conveniently, if we go back to add in those plate appearances, we find that the numbers do, in fact, get pulled back towards league average! Those blip years like 2007? That’s called regression to the mean. It happens in positive ways too. It does.

Stats aren’t magical. In every situation, they predict that the player will most likely be average in all the areas that they haven’t been proven to control. But since he can’t control it, those numbers only work out over time. In any one year, or in this case 500 PA’s, it’s completely up to chance. It’s only once we get to five or six years running with this stuff when it’s viable to pause and say “Well, hm, maybe he’s got a bigger split than we thought before.”

As for Lowrie, again, my problem with sample size comes with things like splits. Not for everything. Though I readily admit that Jed Lowrie’s 2010 numbers could be completely a matter of sample size, and always add that I don’t expect him to come within 50 OPS point of his 900-or-so from last year, it’s just a completely different thing.

It’s overall performance, and we don’t have the numbers we do on Crawford to develop the arguments. It’s basically just a matter of your guess is as good as mine to determine whether he’ll keep it up. I just think there’s more signs that he’s a reasonably better player than Scutaro than there are that he’s not, and that to pass up on the chance for that kind of player when his floor is probably around Scutaro is insane.

Anyways, I feel like the argument has gotten muddled at this point, so to clear up

I state the following:

1. A player does not develop the ability to hit left handed and right handed pitching independently. Actual production against either one is best predicted by applying that player’s splits to the hitter’s overall ability.

2. The difference between a player’s right-handed and left-handed splits cannot be determined by a sample size below 1,000 plate appearances against same-handed pitching.

3. Carl Crawford, over those 1,000 plate appearances, has shown lefty splits about 30 points of OBP and 90 points of SLG below his righty splits.

4. Therefore it is reasonable to expect, should he perform at the same overall level next year as he did in 2010, that Carl Crawford will hit about .330/.420 against lefties.

Finally, what I find interesting is the way you cherry pick stats to show Crawford’s improvement.

?

We know platoon splits stabilize around 1000 PAs. Forgetting everything else, Ben looked at the last 1,000 PA’s more or less. This seems a much more logical approach than your year to year and 3 year splits. Choosing 3 years randomly seems much more akin to cherry picking.

If CC was truly improving, you’d expect to see better numbers; but we don’t.

No. You wouldn’t necessarily. That’s the point. There is simply too much noise and random variation in a single year or even 3 years of platoon split slash stats. Too many random groupings can occur that won’t give us a true reading on that player’s ability.

This isn’t to say Crawford definitely has improved – its to say we don’t know if he has, but given his improvement as a hitter overall (for which we have much more reliable numbers/sample sizes) there is a case to be made that he has.

We would not expect that improvement would show up in Crawford’s stats the way you are parsing them. Its possible it would, but also quite likely it wouldn’t. You simply can’t draw conclusions from ~150-200 PAs a year against LHPs. Its kind of like looking at 3 months of UZR and saying a fielding improvement would have shown up in that on a month to month basis.

Crawford has only had two seasons with an OPS of .830 or better: 2006 (.830) and 2010 (.851). In ’06, his OPS against lefties was .776. Last year it was .690.