[Malleus Dei]

Originally posted by lynchjm24
I love James - and I'm going to try to dig up this work - but doesn't he have to be able to predict the future a bit with this analysis?

Yes, and this is a valid criticism. Suppose that it turns out in 2020 that the best manager of all time was Bob Brenly, now in his second year (I doubt that this will happen, but it could). That would make Brenly the "best current manager" but that is unknowable at this point in time. So James, just as he does to pick the best current players in baseball, uses the best model he can develop along with past performance data to make his rankings.

To give an example of what you are alluding to, suppose that Mike Ryan of the Minnesota Twins, the guy with the two-hits-and-two-runs-and-two-RBI-in-one-inning debut the other day that was erased by the rainstorm, is actually not only the best left fielder in the game today, but the best left fielder in the history of the game. No one would know this, and no one could know this at this point in time. No matter who ranked left fielders, or how they did it, Ryan would not rank first. But he might actually be not only the best current left fielder in the game but also the best left fielder in the history of the game - or he might be one of the worst. We can't know what he is until he shows us on the field. This is why we have to use real data in our models.