the last couple of off-seasons i've run a little exercise using projection data. i aggregate the individual player projections into team totals, then adjust the numbers based on likely playing time. the end result is a team run-scoring projection. in 2006 it was spectacularly accurate --- the exercise yielded an estimate of 780 runs, and the team actually scored 781. last year, not so lucky --- the exercise once again forecast about 780 runs, but the cardinals only scored 725; missed by 7.5 percent.
i normally don't do this projection until the roster is pretty well set, and the 2008 roster may very well change; the cards might still add a hitter, or then again they might trade rolen. so it's premature to be conducting this exercise. but with edmonds and eckstein gone and the cards moving into full-blown rebuilding mode, i got curious about where they stand as of this moment. so i decided to take a preliminary look, based on the roster currently in hand.
let's dispense up front with the obligatory warnings about the hazards of using projection data (i'll be working with ZIPS and CHONE today). whenever i invoke projection data, somebody inevitably points out in the comments section that these systems often get it wrong (as if we needed someone to tell us) and argues that, therefore, the whole exercise is pointless. it's true that the projection systems are often wrong --- about 30 percent wrong, empirically. but that means they're 70 percent accurate, hence far from pointless. every big-league front office runs a version of this same exercise --- a more sophisticated version, using better projection data and more accurate assumptions about player health and playing time --- to gauge their rosters' strengths and weaknesses. if teams are allocating big chunks of time and money to this sort of thing, then i guess there's no harm in our devoting an occasional post to it.
so here we go. i constructed this roster out of the following core players:
infielders: pujols, kennedy, ryan, izturis, rolen, spiezio
outfielders: duncan, ankiel, ludwick, schumaker, barton, rasmus
catchers: molina, larue
for each guy, i plugged in dan szymbroski's ZIPS data (which came out back in october) and pro-rated it for likely playing time, so that my team ended up with a realistic number of at-bats position by position; e.g., i assumed about 1900 at-bats for the outfielders (they took 1905 last season), about 1270 for the middle infielders (1245 last year), etc etc. i tossed in a generic batting line for pitchers and another for replacement players (the minor-league callups and dfa acquisitions who move on and off the roster every summer) and ended up with a full season's worth of at-bats. here's my chart:
AB | H | 2B | 3B | HR | W | AVG | OBP | SLG | BR | |||
pujols | 545 | 177 | 33 | 1 | 37 | 103 | .325 | .432 | .593 | 130 | ||
ryan | 500 | 133 | 23 | 3 | 5 | 37 | .265 | .315 | .351 | 53 | ||
rolen | 480 | 123 | 32 | 1 | 11 | 54 | .255 | .331 | .387 | 62 | ||
duncan | 455 | 119 | 23 | 1 | 23 | 59 | .262 | .346 | .472 | 72 | ||
ankiel | 450 | 108 | 18 | 2 | 25 | 27 | .241 | .284 | .458 | 59 | ||
kennedy | 440 | 110 | 24 | 2 | 2 | 37 | .251 | .309 | .332 | 43 | ||
ludwick | 400 | 105 | 25 | 1 | 19 | 41 | .262 | .330 | .476 | 61 | ||
molina | 400 | 98 | 21 | 0 | 7 | 36 | .245 | .307 | .349 | 42 | ||
itzturis | 330 | 83 | 17 | 1 | 1 | 21 | .252 | .296 | .318 | 29 | ||
schumaker | 240 | 65 | 12 | 1 | 3 | 15 | .272 | .314 | .370 | 26 | ||
spiezio | 200 | 51 | 10 | 1 | 5 | 29 | .253 | .347 | .392 | 27 | ||
larue | 175 | 36 | 8 | 1 | 5 | 24 | .206 | .301 | .354 | 19 | ||
rasmus | 150 | 37 | 8 | 1 | 5 | 19 | .245 | .328 | .419 | 20 | ||
barton | 150 | 37 | 7 | 1 | 3 | 18 | .245 | .325 | .357 | 17 | ||
repl players | 340 | 88 | 21 | 2 | 5 | 32 | .259 | .322 | .376 | 40 | ||
pitchers | 300 | 45 | 6 | 0 | 2 | 15 | .150 | .177 | .190 | 10 | ||
PROJ TOTAL | 5555 | 1414 | 288 | 19 | 160 | 560 | .255 | .323 | .400 | 701 | ||
2007 TOTAL | 5520 | 1513 | 279 | 13 | 141 | 560 | .274 | .340 | .405 | 736 |
that big bold red number is this roster's projected output over 162 games --- 701 runs, using the base runs estimation model, for those of you scoring at home. the actual 2007 team totals are provided for context; the st louis offense projects to be about 35 runs worse than last year's, a dip of almost 5 percent.
ouch.
basically, ZIPS doesn't think the cardinals can maintain the high batting average that kept them near the middle of the pack last year. insofar as the cards let go of some high-average players (eckstein, taguchi, and miles to name 3), that's not an unreasonable conclusion for ZIPS to draw. the system also doesn't think either ankiel or ryan will hit as well over a full season as he did in a small sample last season; again, not unreasonable. it assumes molina's hitting will regress to its former level of crapitude and that rolen's will remain at its current level of crapitude; i think ZIPS is wrong on both of those counts. but on the whole, ZIPS thinks the cardinals have an atrocious offensive team; 701 runs would have ranked 14th in the nl last year, ahead of only the giants and nationals.
i ran the same numbers using CHONE projection data and got a result that isn't quite as dire:
AB | H | 2B | 3B | HR | W | AVG | OBP | SLG | BR | |||
ZIPS | 5555 | 1414 | 288 | 19 | 160 | 560 | .255 | .323 | .400 | 701 | ||
CHONE | 5555 | 1439 | 280 | 23 | 171 | 589 | .259 | .330 | .410 | 733 | ||
2007 TOTAL | 5520 | 1513 | 279 | 13 | 141 | 560 | .274 | .340 | .405 | 736 |
CHONE is considerably sunnier than ZIPS on rolen, and slightly more so vis-vis kennedy and ankiel; it even figures cesar izturis will manage a .320 on-base pct. per this sytem (or my distillation of it anyway), the st louis offense has gotten neither better nor worse. it wasn't very good to begin with, so maintaining the status quo is nothing to brag about, but at least we're not talking about finishing dead last in the league in runs.
if there's anything encouraging to be taken from these estimates, it's that both systems figure the cards will hit for more power than they did last season. but by and large, these cocktail-napkin figures reinforce our gut impressions of this offense: it ain't very good.