Player Projection
Dan Duke, at Mariner Central, raises a question that is of the first order of magnitude:
Given that most of us aren't GMs, do you have a preferred way to find a "jumping off point" for a player's expected contribution? For example, the PECOTA weighted mean expectation? Or perhaps the weighted mean with a risk adjustment? How would you go about projecting a player for next year?
...............
I think if you're a GM, you have the responsibility to make your own projections. And, that is what GM's in fact do.
If a major-league General Manager ever used PECOTA's 50th-percentile projections to make all of his decisions during a season, he would truly deserve to be fired for incompetence.
What, Jack Zduriencik isn't down by the batting cage assessing Bill Hall's ability to fix his problems? Zduriencik isn't in the video room, assessing Brandon Morrow's ability to refine his release point? Zduriencik is no better-informed on his own players than Baseball Prospectus is?
................
You play roto, right, Dan? How do you do it? :- )
...............
For bulk listings, I
- (a) start with Ron Shandler's projections;
- (b) look at his SKILL data to find things out of alignment;
- (c) try to consider the RotoTimes player notes and the player's template and comps;
- (d) come up with my own final, um, guess.
Dan asked about a "jumping-off point." You could use any reasonable jumping-off point -- Shandler, PECOTA, whatever. The key is that you then apply your own judgment as to whether you expect the 2010 player to perform better or worse than the generic, vanilla averages.
I use Shandler rather than unadjusted, weighted averages because Shandler has already taken the weighted averages and tweaked them to match things that are out of alignment in the stats -- and he has explained to you the tweaks that he has made.
Just for instance, in 2009, Shandler predicted Kenji Johjima for a rebound, for his SLG to go from .336 actual in 2008 to .415 projected in 2009 (Joh actually slugged .406). He explained why: "age, xAVG, H% and history." In other words, Joh had a lousy BABIP in 2008; he is still relatively young; and he has proven himself a quality hitter.
Johjima's 2008 H% was out of alignment, and spotting that, Shandler correctly predicted him to rebound. Any Mariners fan would have, but Shandler isn't a Mariners fan.
..........
Some players we have time to project on a more detailed level. For the Mariners, or guys we POTD like J.J. Hardy, we reason it out as detailed in the POTD's on SSI. My reasons for projecting Hardy are the same as those listed: declining LD rate, poor FB performance ... small, weak RH "power" hitter template, etc.
When Tim Lincecum and Jered Weaver and Justin Verlander were in AAA, I was projecting them to make instant star impact in the major leagues based not on a weighted-mean MLE formula, but based on the factors that are spelled out in the Lincecum POTD's. Such as the fact that Lincecum's giant pinwheel-leap off the mound, and hidden 55-foot fastball, were backed by shocking results: Timmy had twice the K's (and twice the wheat) that Brand Morrow did in the same college league.
Not sure if that addresses what you're talking about. :- )
..................
One thing I fervently avoid doing, is letting a weighted-mean formula dictate to me the "correct" projection for a player. I will guarantee you that Jack Zduriencik never finishes the cabinet with such a naive sanding process.
I respect a roto enemy who thinks that he can win a league with (say) PECOTA or some other automated projection system.
I've got no respect for a paradigm that says automated projections give us "correct" evaluations and that deviating from them is "wrong." Anybody who wants to publicly commit to using PECOTA-only decisionmaking processes, is quite welcome to finish 10th of 12 against us in next year's roto league. ;- )
The GMs who win World Series, such as Gillick and Epstein, blend "automated" projections with their own advanced intel, and superior intuition, as to who is going to have a good year.
...........
As Bill James told us recently vis-a-vis the way the Red Sox do it, "You just have to look at all of it. You have to look at the [computer] projections, and what's out of alignment in his [skills] trends, and where a guy's FB velocity is going, and what the scouting reports are, and what the doctors are saying, and how his new pitch is coming along, and you have to make a call."
We don't want to buy into the idea that weighted-mean averages are good enough. That would be naive.
Cheers,
Dr D