Last year, we received plenty of good feedback on my attempt to project the 2007 Red Sox and Yankees. As I've described earlier this offseason, one of the methods I used came up with a remarkable 105 wins -- four off the Red Sox' 2007 Pythag record. That was a flawed prediction, and I won't be repeating the method for reasons I'll detail below. But the other, more accurate method came up with 99 wins -- just three off the Sox' actual total and two off the Pythag -- basically splitting the difference. Not bad.
Since you all seemed to enjoy it -- and since the projections didn't fall flat on their faces -- here is Part 1 of our second annual Predictions Special, projecting the Red Sox. The Yankees' projections will be forthcoming.
Last year, I started by using a hybrid of Bill James and PECOTA, but determined that the difference between the individual projections was negligible on offense, and James was far too optimistic in projecting pitchers. This year, let's try only PECOTA, which has been determined time and again to be the most accurate of the projection systems out there anyway.
First, a look back at the Sox' 2007 lineup:
|Pedroia||.317/.380/.442, 8 HR, 48 XBH, 47 BB, 42 K, 86 R|
|Youkilis||.288/.390/.453, 16 HR, 53 XBH, 77 BB, 83 RBI|
|Ortiz||.332/.445/.621, 35 HR, 88 XBH, 111 BB, 117 RBI|
|Ramirez||.296/.388/.493, 20 HR, 54 XBH, 71 BB, 88 RBI|
|Lowell||.324/.378/.501, 21 HR, 60 XBH, 191 H, 120 RBI|
|Drew||.270/.373/.423, 11 HR, 45 XBH, 79 BB, 64 RBI|
|Varitek||.255/.367/.421, 17 HR, 35 XBH, 122 K, 68 RBI|
|Crisp||.268/.330/.382, 6 HR, 41 XBH, 60 RBI, 28 SB|
|Lugo||.237/.294/.349, 8 HR, 46 XBH, 73 RBI, 33 SB|
This was your typical lineup -- Lowell and Drew flipped back and forth, Crisp was in and out as Ellsbury heated up, Manny missed a month, etc. But on any given day in 2007, this was the lineup trotted out there -- and it was largely a top-heavy beast, a gaping hole beyond Mike Lowell in the five spot until Drew got hot in August.
Using the Lineup Analysis Tool, that lineup should have scored 5.80 runs per game, or 940 runs. The best iteration could have scored 5.94 (if you want to bat Ortiz second and Drew ninth, with Lowell at cleanup). The worst would score 5.48 (with Lugo leading off). In fact, the Red Sox last year scored 867 runs, or 5.35 per game -- a loss of about seven wins. That's what giving Wily Mo Pena, Eric Hinske and Alex Cora a season's worth of at-bats will do for you.
How does that compare to PECOTA's projections for 2008?
|Ellsbury||.287/.346/.395, 5 HR, 41 XBH, 52 RBI, 32 SB|
|Pedroia||.295/.361/.430, 10 HR, 51 XBH, 53 BB, 50 K, 81 R|
|Ortiz||.282/.402/.541, 35 HR, 74 XBH, 111 BB, 119 RBI|
|Ramirez||.281/.381/.486, 21 HR, 51 XBH, 72 BB, 84 RBI|
|Youkilis||.272/.373/.448, 17 HR, 55 XBH, 79 BB, 81 RBI|
|Lowell||.285/.345/.437, 14 HR, 50 XBH, 146 H, 79 RBI|
|Drew||.267/.346/.421, 11 HR, 39 XBH, 65 BB, 60 RBI|
|Varitek||.255/.352/.420, 11 HR, 29 XBH, 86 K, 49 RBI (384 PA)|
|Lugo||.275/.337/.384, 6 HR, 36 XBH, 46 RBI, 21 SB|
Well, that doesn't look as good as I'd have predicted. A continuing fall-off from Drew that would make his contract one of the worst in memory, no bounce-back from Ramirez, regressions from pretty much everyone, with the only improvements coming from Lugo and by Ellsbury replacing Crisp. Such a lineup putting up those OBPs and slugging percentages would score 881 runs (5.44 runs per game) -- marginally better than the Sox' actual 2007, but not nearly as good as the 5.8 the 2007 lineup "should" have scored (which is what really matters, as you have to assume injuries and bench time will remain constant). The best lineup would score 5.52, the worst 5.27. That puts the Sox' 2008 starting lineup at 57 runs worse than 2007's -- nearly six wins.
For comparison's sake, Dave Pinto has looked at Boston's projections using the lineup tool and the Marcel and Bill James systems, and came up with 5.78 RPG in Marcel (essentially no change from 2007), and 5.95 runs per game under the Bill James numbers (a slight increase). It should be noted that Pinto uses Pedroia at leadoff, Youkilis second, Drew fifth and Ellsbury eighth in his lineups, which probably would result in a better-scoring scenario (but not by a significant margin).
So, unlike last year, when PECOTA and James were very close (three runs apart), the projections seem to vary on how to treat the poor years from Ramirez, Drew and Lugo, as well as how much to regress (or progress) youngsters like Pedroia and Ellsbury. The Sox could gain two or three wins (Bill James), see no change (Marcel) or lose six wins (PECOTA) on offense. Averaging the three gets us 931 runs, about one win down.
Now, let's take a look at the pitching staff, obviously the Sox' bright spot -- it's fairly young, getting younger and was the best in the league last year. Could it possibly improve?
2007 stats (ERA/WHIP/BAA)
|Beckett||20-7, 3.27/1.141/.239, 200.2 IP, 194 K, 40 BB|
|Matsuzaka||15-12, 4.40/1.324/.237, 204.2 IP, 201 K, 80 BB|
|Schilling||9-8, 3.87/1.245/.267, 151 IP, 101 K, 23 BB|
|Wakefield||17-12, 4.76/1.349/.252, 189 IP, 110 K, 64 BB|
|Tavarez||7-11, 5.15/1.500/.272, 134.2 IP, 77 K, 51 BB|
|Lester||4-0, 4.57/1.460/.244, 63 IP, 50 K, 31 BB|
And the top four bullpen arms:
|Papelbon||1-3, 1.85/0.771/.146, 58.1 IP, 84 K, 15 BB|
|Okajima||3-2, 2.22/0.971/.242, 69 IP, 63 K, 17 BB|
|Timlin||2-1, 3.42/1.084/.217, 55.1 IP, 31 K, 14 BB|
|Delcarmen||0-0, 2.05/1.023/.175, 44 IP, 41 K, 17 BB|
The Red Sox allowed 657 runs in about 1,439 innings, or 4.06 runs per game. It received a Cy Young-caliber performance from Beckett, decent seasons from Matsuzaka and Schilling, more of the same from Wakefield, and an overall excellent line from Tavarez and Lester, considering they were the fifth pitchers. The bullpen, of course, was amazing, adding a solid performance from Kyle Snyder with the four excellent ones listed.
This year, one assumes Lester will be in the fifth spot full-time, splitting innings at some point with Clay Buchholz. Here's what that rotation would look like:
|Beckett||14-8, 3.64/1.22/.237, 205 IP, 176 K, 60 BB|
|Matsuzaka||13-8, 3.90/1.29/.238, 191 IP, 170 K, 67 BB|
|Schilling||8-6, 4.18/1.29/.264, 123.1 IP, 87 K, 26 BB|
|Wakefield||8-7, 4.86/1.44/.263, 131 IP, 76 K, 49 BB|
|Lester||8-8, 4.93/1.56/.259, 128.2 IP, 92 K, 65 BB|
|Buchholz||8-7, 4.08/1.35/.235, 130.2 IP, 119 K, 56 BB|
The uptake of these results is that the sooner the Red Sox get Clay Buchholz into the rotation, the better. PECOTA thinks he'll be better than Curt Schilling this year, putting up K/9 numbers on par with Beckett and Matsuzaka, who is projected to have a very nice sophomore season (eighth among AL starters in projected ERA). In fact, these projections have four Sox starters in the top 16 in ERA for 2008. So not bad at all.
The top four bullpen arms:
|Papelbon||5-5, 2.55/1.06/.202, 67.1 IP, 81 K, 21 BB|
|Okajima||3-3, 3.71/1.29/.242, 56.1 IP, 50 K, 19 BB|
|Delcarmen||3-2, 3.81/1.34/.234, 58.2 IP, 53 K, 25 BB|
|Aardsma||3-2, 4.22/1.38/.238, 48 IP, 42 K, 21 BB|
Hello, David Aardsma, beneficiary of the fact that PECOTA isn't a big fan of Timlin's chances in 2008 (4.46/1.36). Papelbon's ERA is the lowest of the 976 pitchers projected for 2008. Semirelatedly, the ZiPS system set a record with its ultralow projection for Papelbon's 2008 ERA, as well. PECOTA, conservative as ever, sees Okajima as being the real deal, another step forward from Delcarmen, and Aardmsa being a key part of the pen. It's obvious that if the '08 Sox have any holes, they're not among the hurlers.
Projecting those projections (heh) over the 1,439 innings thrown by Sox hurlers in 2007 results in 648 runs allowed -- nine fewer than 2007, for a whopping one-win gain.
One of the biggest critiques brought up by commenters last year was that this method assumes the entire pitching staff will perform at the level of its top 10 or 11 arms. Figuring that was valid (after all, the starting nine over 162 games clearly produces a lineup about seven or eight wins better than what actually occurs when you include the bench), I adjusted the numbers for the 11 pitchers I profiled last season, giving them all 1,439 of the team's innings to see how much less scoring they allowed. Hopefully, I reasoned, that would give us an idea how much to adjust this year's projections. The answer? They would have allowed one fewer run. That may be a fluke caused by the superior backup pitchers the Sox had (Gabbard, Buchholz, Snyder and Romero all had very low ERAs), but I think it allows us to trust that the staff projections aren't affected as much as one might think by the 300 innings or so thrown by other pitchers whose use we can't foresee.
On offense, the dropoff caused by the use of the bench is included by focusing only on how much the lineup should score from year to year and determining runs gained or lost solely on the starting nine. That tells us whether the starters will perform better (presumably having the most effect), and we can assume that 70 or so fewer runs will be scored by using bench players.
Last year, I provided a straight prediction based on the Pythag record produced by the projections, without a comparison to the previous season. That's what produced the 105-win prediction. I'll pass on doing that this year, as it clearly fails to take into account a pretty significant drop from the starters to the reserves on offense.
To recap, here's a chart giving last year's lineup "shoulda-scored" runs for the starting nine, this year's projected runs for the top nine, last year's actual runs allowed and this year's projected runs allowed based on the 10 pitchers' projections I posted above:
Talk about consistency. The Sox are projected to score nine fewer runs -- but give up nine fewer runs, as well. On the surface, that looks like we should expect another 96-66 season, but not so fast.
This is based on run differential, after all. And last year, the Red Sox were, believe it or not, an unlucky team, going 22-28 in one-run games and finishing five games behind their Pythagorean record. Based on their run differential in 2007, the Red Sox should have finished 101-61.
So there you have it. The projections for 2008 say the Red Sox should be a 100-win team -- just like they should have been in 2007. The key, in both halves of that sentence, is should.