team projections: PECOTA

last week we aggregated the cardinals' ZIPS projections to derive team projections for the hitters and the pitchers. today we'll do the same thing with the PECOTA projections. as i mentioned on tuesday, PECOTAs are proprietary, so it's not cool to publish a full team's worth of them; accordingly i will be publishing the cumulative numbers only today, without the accompanying player-by-player breakdowns.

before i get to those team projections, here's some cool but useless PECOTA trivia:

• how highly does PECOTA think of albert pujols? he ranks first in all of baseball in projected batting average, slugging, runs scored, equivalent average, VORP (or value over replacement player), and WARP (wins over replacement). he ranks 2d in on-base percentage, behind barry bonds; he's 3d in rbi.
• carpenter's four top comparables are 1 gaylord perry, 2 roger clemens, 3 rick reuschel, and 4 orel hershiser. reuschel seems like the odd man out here; he had a good career, won 200+ games, but was less special than the other three.
• among pitchers who are projected to make 20 or more starts, st louis has 3 of the 11 lowest projected eras in the national league: carpenter, reyes, and wainwright. the top 15 are: 1 sheets, 2 peavy, 3 carpenter, 4 oswalt, 5 webb, 6 smoltz, 7 zambrano, 8 hamels, 9 maddux, 10 an reyes, 11 wainwright, 12 glavine, 13 chr young, 14 dontrelle, 15 schmidt. observations: the padres have 3 of the top 13 (peavey, maddux, and young); also, per this list all three anchors of the late great atlanta braves' rotation remain among the nl's best starters; the trio were last teammates in 2002, and haven't been together in the same rotation since 1999, yet if you got them together today they'd still constitute a formidable staff.
• continuing to look at rotations: the cards' top three starters project to rack up a total of 13.2 wins over replacement, which ranks 4th in the league and second in the nl central; the brewers' top 3 starters (sheets, capuano, and bush) have the highest combined WARP in the nl, at 13.6. given the card threesome's low eras, why don't they rank first? simple answer: reyes and wainwright only project to throw about 160 innings apiece, which reduces their projected impact on the win column. the top 4 are milwaukee, the dbacks (with webb and big unit), philadelphia, and st louis.

ok, so much for that; on to the feature presentation. without further ado, here's roughly how PECOTA sees the cardinal offense performing at the team level; ZIPS projections and actual 2006 stats included for context:

			AB	H	2B	3B	HR	BB	|	AVG	OBP	SLG	|	runs created
PECOTA			5540	1464	284	31	172	564	|	.264	.338	.420	|	788
ZIPS			5540	1468	289	20	172	551	|	.265	.338	.419	|	780
2006			5552	1484	292	27	184	531	|	.269	.337	.431	|	804

remember, last year's actual runs-scored total was 781, not 804; the latter is merely the number of runs we would have predicted based on the cards' distribution of hits, walks, and extra bases. so PECOTA projects the team to stay at about the same level offensively, and it's ever so slightly more bullish than ZIPS on the st louis offense -- by roughly one win. the distribution of at-bats i used in this exercise is essentially the same as the one used in the ZIPS team projection; refer back to that table if you want to see the breakdown. the only real difference between ZIPS and PECOTA is in the triples category; PECOTA, for some reason, has the cards hitting 11 three-baggers more, a totally meaningless blip. in the important categories --- homers, avg, obp, slg --- these two systems are almost entirely in synch.

i'll look at the pitching staff after the jump.

things have changed rather dramatically since we ran the ZIPS-based projection last week. the cardinals signed two starting pitchers, mulder and franklin; accordingly, we'll need to adjust our assumptions about playing time. because ZIPS figures are freely distributed, i can use them to illustrate how i see things breaking down now. in the original exercise, i had brad thompson as the 5th starter, making 28 starts. let's move him back to the `pen, and assign some starts to the two new signees:

			GS	IP	H	ER	BB	SO	HR	|	ERA	WHIP	|	total runs
carpenter			30	209	187	71	42	170	21	|	3.06	1.096
reyes			28	166	158	71	42	140	26	|	3.85	1.208
wainwright			28	165	174	71	53	102	22	|	3.87	1.378
wells			28	154	154	85	85	116	23	|	4.97	1.554
franklin			22	131	144	73	38	60	20	|	5.02	1.389
mulder			14	80	85	41	30	46	9	|	4.61	1.438
repl starter			12	60	70	38	30	33	12	|	5.70	1.667
is'hausen			0	61	50	24	30	56	6	|	3.54	1.311
looper			0	60	61	24	19	35	4	|	3.60	1.333
thompson			0	62	62	25	17	38	7	|	3.63	1.274
springer			0	50	46	23	18	40	7	|	4.14	1.280
kinney			0	60	55	27	29	47	6	|	4.05	1.400
hancock			0	60	58	27	18	38	7	|	4.05	1.267
flores			0	42	43	21	18	35	4	|	4.50	1.452
t johnson			0	45	40	26	35	47	6	|	5.20	1.667
repl relievrs			0	35	41	20	18	25	7	|	5.14	1.686
new ZIPS			162	1440	1428	666	521	1028	184	|	4.16	1.354	|	722
old ZIPS			162	1441	1431	667	546	1054	190	|	4.17	1.371	|	723
2006			161	1430	1475	721	504	970	193	|	4.54	1.383	|	762

mulder's raw ZIPS projection calls for 27 starts; i simply cut that in half. originally i also put franklin at 14 starts, with the remaining 20 starts on the schedule assigned to a generic replacement starter. but then i decided franklin is a generic replacement starter --- if the cardinals need an emergency fill-in, he'll be the first guy they go to --- so i shifted some of those assignments onto his line. his unadjusted ZIPS projection calls for 22 starts; that's just what i gave him. then i stuck thompson back in the bullpen and pro-rated the other relievers' innings . . . . and ended up about back where i started. the new ZIPS projection is almost identical to the old; mulder/franklin is more or less a wash with brad thompson.

i applied the same distribution of starts to the team PECOTA projection. here's how it came out:

			GS	IP	H	ER	BB	SO	HR	|	ERA	WHIP	|	total runs
PECOTA			162	1440	1460	681	534	1049	178	|	4.26	1.385	|	737
new ZIPS			162	1440	1428	666	521	1028	184	|	4.16	1.354	|	722
2006			161	1430	1475	721	504	970	193	|	4.54	1.383	|	762

yeesh. . . . not encouraging. but what happened? PECOTA loves the top of the st louis rotation --- 3 of the best 11 eras in the league, remember? so why is the overall team projection so much worse per PECOTA than per ZIPS? surprisingly enough, it's the bullpen. here's how the units compare:

			GS	IP	H	ER	BB	SO	HR	|	ERA	WHIP	|	total runs
PECOTA rotation			162	940	961	446	325	667	118	|	4.27	1.368	|	482
ZIPS rotation			162	965	972	450	320	667	129	|	4.20	1.340	|	487
PECOTA pen			0	500	499	235	209	372	60	|	4.23	1.416	|	254
ZIPS pen			0	475	456	217	202	361	54	|	4.11	1.385	|	235

like ZIPS, PECOTA projects the cardinal rotation as considerably better than league average; only 3 nl starting corps had eras lower than 4.27 last season. but the projected 4.30 bullpen era would have ranked 9th in the league last year. . . . . and PECOTA is assigning the 'pen a larger number of innings than ZIPS.

i'm thinking right along with some of you: if the pen's in that bad of shape, suppose the cards could get jeff weaver on their terms (ie, a two-year deal) and sent wainwright back to shore up the relief staff? here's that projection:

			GS	IP	H	ER	BB	SO	HR	|	ERA	WHIP	|	total runs
OVERALL			162	1440	1463	679	524	1047	181	|	4.24	1.381	|	735
rotation			162	941	975	454	319	653	122	|	4.34	1.376	|	490
bullpen			0	499	488	225	205	394	59	|	4.23	1.416	|	244

PECOTA doesn't project wainwright as a reliever, but we do have a ZIPS projection for him in that role, so i used it; that lowers his era by two-thirds of a run. to fit adam into the bullpen, i simply kicked braden looper off the team; he has the worst PECOTA projection among the relievers. these two maneuvers optimized the marginal gain in the bullpen --- a 9-run improvement, or one win in the standings. and it's canceled out by the 8-run hit the rotation takes with the addition of weaver, whose PECOTA era is nearly half a run higher than wainwright's. so if a weaver signing precipitates the return of wagonmaker to the bullpen, it's a useless idea --- per these assumptions, anyway. worse than useless, since it interferes with the organizational imperative to develop young starting pitchers.

but suppose they signed weaver and bumped franklin from the rotation, while wagonmaker stayed there? different story:

			GS	IP	H	ER	BB	SO	HR	|	ERA	WHIP	|	total runs
PECOTA 1			162	1440	1460	681	534	1049	178	|	4.26	1.385	|	737
PECOTA 2			162	1440	1463	679	524	1047	181	|	4.24	1.381	|	735
PECOTA 3			162	1440	1440	669	514	1061	171	|	4.18	1.356	|	725

just to refresh your memory here, "PECOTA 1" is the weaverless projection; franklin takes 22 starts, mulder 14, etc etc. in "PECOTA 2," weaver replaces wainwright in the rotation, and wainwright replaces looper in the bullpen. under "PECOTA 3," weaver signs and ryan franklin doesn't make the team. the great improvement under this scenario lies in the rotation's stability; in PECOTA 3, carpenter takes 30 starts, reyes wainwright wells and weaver get 28 apiece, and mulder gets 14, which means we only have to assign 6 to our generic replacement starter --- the stiff with the 5.75 era. the rotation's collective era drops to 4.16, and the bullpen gets assigned a smaller number of innings; both of those things drive the overall team era downward.

it's still only a 1-game improvement, though; nothing to break the bank over. jocketty should continue to hold the line with boras and offer weaver two years, take it or leave it; maybe add an optional 3d year. if jeff agrees to those terms --- and the manager optimizes the use of the roster resources --- a weaver signing might be worth the trouble.

if anybody's still reading, it's time to add this all up. we run the pythagorean formula on totals of 788 runs scored and 737 runs allowed (ie, the PECOTA 1 projection) to arrive at a projected won-loss record of (drum rollllllllllll. . . . . . . . ) 86-76. if we take the PECOTA 3 projection (weaver's back!), the projected record is 88-74. in truth, the difference is only 1.28 wins; we get 86.41 wins under the first scenario, and 87.73 wins under the second, so rounding off creates the illusion of a 2-win spread. the upshot: PECOTA has the cardinals right where ZIPS did, with a win total in the mid/high 80s.

i feel like i can trust those projections, as they're based on some rather conservative assumptions re playing time. the biggest "if" is the one we've been debating back and forth since november: how reliable are those young starting pitchers? if reyes and wainwright hit their projections, the team's going to contend; if they don't, the team is in trouble. it's almost that simple.