Fri Nov 09, 2018 2:51 pm

Thanks for the comments. I guess I'm still not sure.

If it's as Steve just indicated, perhaps it shouldn't be employed at all in the online game. Not because of any fairness issue, since I've always assumed it gets built into the pricing model, but because it is based on a measure which doesn't even translate into the circumstance in which it is employed.

If it's consistent with the windows game help file, then why not devise the rule to apply more subtly across more circumstances? And let pitchers have a tougher (or weaker) with-runners-on rating too. Bigger sample size = reduced statistical variation, right? There are, after all, plenty of cases where a pitcher with a relatively high WHIP managed an ERA which was lower than you'd expect, because he was tough in jams (or maybe tough and lucky).

The Strat statement about the problem with small sample size is a little silly, considering there are cards for players with fewer than 100 at bats, which themselves have clutch ratings. Talk about a small sample size! Take 2001 Ed Sprague, with 105 PA's. He's got one clutch out chance on his card, each side. That's a one in 216 outcome. Being generous and assuming 20% of his at bats come in clutch situations, and assuming you use him and limit use to real-life, that comes up once every 10 full season simulations, on average.

It seems to me the cards are meant to simulate the production of a player in a specific season. The cards don't "accurately measure" every player's clutch "abilities", since there are many players who were either clutch or not over their careers who had years where their stats skewed the other way.

I think the leverage measure is an interesting one, distinguishing outcomes in at bats with more or less potential impact on the game outcome.