Now, I would say the "Mookie only hit XXXX against lefties!" is not a good argument against the move. 60-game platoon splits for a hitter are nearly worthless from a predictive standpoint.

If all you knew about a hitter was 246 PA, I have generic platoon splits as ~93% of model

Mookie Betts is only at the point in his career in which his actual platoon splits (-12 point advantage against lefties) are about *equally* as predictive for future platoon splits as a generic righty's platoon splits.

We have enough of Betts that we should only be predicting a 15-20 points-of-OPS advantage against lefties rather than a more normal 40-50 point one.

The argument for keeping Snell in (which I favor!) has little to do with Betts platoon splits.

The argument for keeping Snell in is a much simpler one: He was pitching well, efficiently, styming the Dodgers, and he's not starting again for at least five months.

No reason to complicate things unnecessarily!

I know it might sound like an odd thing for a notorious number-cruncher like me to say, but if you've read my work, you may be surprised how few stats I actually use in direct writing.

Stats guide an intellectually honest story, but they're not typically the story themselves!

Unless we're talking fun stuff like records or .400 BA things, stats are far more important for what they *mean* than what they *are*.

And analytics largely help you reach *an* answer, not *the* answer. As long as we're not talking crazy stuff like giving Ian Desmond a five-year contract to play first base/corner outfield, they don't usually *demand* a course of action.