2d installment in valatan's all-time card'l series is up --- 2bmen.
last spring, john sickels launched an experiment at Minor League Ball: community projections. he asked every reader to project a stat line -- avg, obp, slg, and steals -- for corey patterson, using nothing more than common sense and intuition to guide them. sickels specifically forbade his readers from relying on PECOTA, ZIPS, and other advanced forecasting methods; he wanted to know how an unscientific projection would compare to a statistically modeled one -- and even whether it might prove to be more accurate.
sickels got 64 responses on corey patterson and averaged them out. here was the community's aggregate projection:
| avg | obp | slg | sb |
|---|---|---|---|
| .275 | .333 | .475 | 27 |
not very accurate, as you're no doubt aware; patterson actually posted a .215/.254/.348 line with 15 sb. but the community's projection was no further off the mark than the "professional" forecasts published by baseball prospectus, bill james, ron shandler, and other sayers of baseball sooth.
by the end of the spring sickels had run community projections on about two dozen players, and the community's lines usually ended up in roughly the same range as the lines generated by the stat models. some were very accurate -- for example, in the case of mark mulder:
| w-l | era | inn | bb/k | |
|---|---|---|---|---|
| projected | 15-9 | 3.86 | 194 | 57/135 |
| actual | 16-8 | 3.64 | 205 | 70/111 |
others were wide of the mark in varying degrees -- but, again, almost always very close to what the expert forecasters came up with.
a couple of other SB Nationeers have picked up the idea. for example, the northsiders at Bleed Cubbie Blue think kerry wood's gonna muster 149 innings and go 11-6 with a 3.79 era and 162 strikeouts. and the oriole partisans at Camden Chat projected corey patterson (the player who started this whole thing) to .261/.299/.432 with 18 homers and 19 steals.
i'm flying out of town for a conference this morning and flying back tonight, and i'll be nowhere near a computer all day -- perfect time to try this idea. if the response is good and people seem to have fun with it, we'll do some more of these; if you think it's a stupid idea say so in the comments, and if that's the consensus i'll retire the idea.
let's use jim edmonds as a test case. he'll turn 36 this year and is coming off his weakest season as a cardinal. do you think he'll bounce back in 2006, or was '05 the beginning of an irreversible decline? ground rules: assume 450 at-bats for jimmy (he has averaged about 475 over the last four seasons) and 550 plate appearances; please project totals in these six categories:
- batting avg
- on-base pct
- slugging avg
- homers
- runs scored
- runs batted in
| avg | obp | slg | hr | rbi | r | |
|---|---|---|---|---|---|---|
| 2002 | .311 | .420 | .561 | 28 | 83 | 96 |
| 2003 | .275 | .385 | .617 | 39 | 89 | 89 |
| 2004 | .301 | .418 | .643 | 42 | 111 | 102 |
| 2005 | .263 | .385 | .533 | 29 | 89 | 88 |
and here are some potentially useful references:
edmonds' 2005 splits at espn
his baseball reference page, including his 10 most similar hitters
his davenport translations at baseball prospectus
post your projections in the comments. you are welcome -- but not obligated -- to include the rationale behind your projection. i'm hoping we can get a minimum of 50 entries; don't be shy, people.
i'll have standard & poor's tally 'em up on friday morning and post the results.