THE BOOK--Playing The Percentages In Baseball

Is the difference in apparent movement on his curveball between ‘07 and ‘08 “real” (i.e. indicative of a true change in skill) or is it demonstrating the effect of Coors field?

Did Coors mask maturation or is Cincy a return to his KC stuff?  The former might point to a real change in true skill while the latter night point to some significant regression in the future.

I can see Dave’s argument but I could also see a GM being wary of going all in on Affeldt. He (Affeldt) seems to be falling in to that grey area that makes life and baseball so much fun. I do agree with those that would suggest it’s not really a compelling argument to suggest Affeldt is one of the most underrated pitchers in baseball though.

You certainly cannot include data from Coors Field in trying to assess whether a player’s fundamental skills, like pitch speed and movement, have changed, unless you at least have a pretty darn good park adjustment for pitch speed and movement.

Josh Kalk’s data does in fact have a pretty darn good park adjustment for pitch speed and movement.  I’d still want to look at road data for Affeldt to make sure.

For what its worth, BIS has four years worth of velocity data published on fangraphs.  Even from his KC days, the 2008 number represents a spike.

I too think that anointing him one one of the best left-handers in the game, closer-quality, and the most underrated player or pitcher in the off-season is a bit presumptuous and premature.

That being said, it is interesting when a player’s underlying skills have appeared to change (skills than we can measure, not some scout telling us that so-and-so is “pitching with more confidence") along with his performance.

Much research needs to be done in this area.  What do we do with that?  Do we simply weight his recent performance more heavily.  Do we regress out weighted average of all his historical performance to a different mean (pitchers with different “stuff” should have different population means, no?)?  Do we do both (probably)?

He is obviously better than we “thought” (his pre-season Marcel), at least we now think he is, as his Marcel has changed (remember that what we “think” about a player is based on a running weighted average of his past performance and thus always changes).  We can also probably add a tick or two since some of his underlying numbers (speed only I guess) have changed.  The question is how much of a tick and in what areas (HR rate, BB rate, K rate, all of the above?).

So, while a player whose performance numbers only go up or down merits a different Marcel than a player whose performance numbers go up or down AND we can measure a change in something more granular (like pitch speed or movement), how much different, I we certainly don’t know.  I wouldn’t think that we should ignore past performance before the “fundamental change” for a variety of reasons.  As well, we certainly still have to regress whatever historical performances we are using, towards some mean.  While David is on the right track here, as I said, I think he overstated his case a little.