Don Mattingly is “making history” in his stint as interim manager of the Phillies, The Athletic excitedly announces today.
Nine and 19 when he took over, the Phils have gone 16-5 since. Impressive!
But can we really infer that his managerial acumen is responsible for turning the team around?
I doubt it.
This is the sort of question that the mWAR Estimator is supposed to help us answer. It uses Bayesian methods to identify the latent or innate skill of a manager. The estimates take the form of wins added per 162 games.
A manager’s mw162 is his Bayesian MAP—or most likely impact. But the estimator actually calculates a mw162 “probability mass,” or probability distribution, which allows us to assess how confident we should be that a manager’s estimated mw162 exceeds any particular threshold one considers important.
In my Journal of Sprots Analytics article, I assess manager mw162s in relation to a “region of practical equivalence” (ROPE) of ± 2, figuring that an impact of that many wins or losses per season should be taken seriously by any practical-minded observer. Here are ROPEs for a select group of managers:
Now here is Mattingly’s ROPE, which reflects his 12 seasons managing first the Dodgers and then the Marlins.
Mattingly’s -0.86 mw162 is decidedly mediocre. It ranks him 399 out of 512 managers (the number who did more than temporary stints at the helm) over AL/NL history.
He’s on tear. Good for him. And of course, we should “update” our manager skill estimate as his current tour of duty continues (it won’t have budged in a bare 21 games).
But don’t be fooled by randomness, as they say. Hot spells like this happen all the time in all manner of baseball performance, and are of essentially zero probative value.
Read the paper if you want to learn more, and if you want to check out all-time rankings, and even do your own analyses like the one I’ve posted here, head over to the data library.