That seems a bit...off. The Red Sox finished in 2006 with an 86-76 record. They scored 820 runs and gave up 825 for a Pythagorean record closer to 81-81. Somehow, this team is going to improve by 21 Pythagorean games while scoring 140 more runs and giving up 110 fewer? I'm a bit skeptical.

It's not that insane, really. PECOTA predicts a rather significant rebound year from Crisp, a decent season from Varitek, and a much improved RF, assuming Drew signs. It's also very much in love with Pedroia, and calls slight declines from Lowell and Ortiz. Manny's supposed to be himself. So basically, it thinks Boston shows a decent improvement in CF, C, RF, 2B, and SS, a slight decline at 3rd and DH, and Manny being Manny in LF.

Further, Boston scored about a half a run less per game (83 total) last year than they "should" have, as noted on that post. Assuming better luck, especially in the injury department, I don't see why Boston can't score a ton more runs this year then last. Maybe not 140 more, but there should still be a dramatic improvement.

As for pitching...Beckett and Schilling's ERA's last year, based on their peripherals, should have been better. Barring injury woes, Beckett will be a far better pitcher this season, and if Schilling doesn't collapse physically, he'll be solid again too. Matsuzaka is surely an improvement on...whoever you want to call last year's No. 3. Same with Papelbon over last season's No. 4. And even if Wakefield does repeat last season's performance...a 4.63 ERA in the 5 spot doesn't look too bad in the AL East. Assuming he's healthy the whole season, of course.

As for the bullpen...that does look like a mess. Really hard to predict right now. But on the strength of that staff, I would guess that it'd have to be a historically pathetic relief corps to avoid a dramatic improvement. They won't win 105 games, but I'd say it's hard to call for them to finish with less then 90-95. Same for NY, and the Jays aren't exactly pushovers anymore...so it'll be an interesting season.

Look at it this way. The Sox are trading off a small amount of defense overall (Gonzalez/Loretta/Crisp/Nixon v. Lugo/Pedroia/Drew/Crisp) for considerably more offense - potentially as many as 100 runs worth. Add in minor rebounds from Varitek and Crisp, and this looks a little optimistic, but not too much so.

On the other side of the ball, it's a consequence of the projection for Matsuzaka, a rebound from Beckett, the removal of Clement and Snyder and the promotion of Papelbon. That also seems not entirely unrealistic.

Will it actually play out that way? Of course not. The best projection systems in the world (and PECOTA is probably one of them) tend to be within a reasonable distance 75% of the time. When they miss, the tendency is to miss high, because it's more likely that someone gets hurt or declines than they suddenly get better. Finally, most projection systems are not good at the margins i.e. projecting two 100 wins teams in the same division, when they have to play each other 19 times, and not taking into account the strength of schedule.

That's ~16 projections each of which could be off by a lot a quarter of the time, and will be tend to be low when it misses. This is the single most likely outcome, but there are a lot more outcomes lower than that.

However, what this does tell you is that the Sox are definitely among the class of the American League again. If the RSox/Yankees/Jays (don't forget them; I'd like to see their projections, too) don't produce the wild card team, it will only be because the Indians/WSox/Tigers/Twins put together an even better division.

The A's/Angels better win their division, because the loser is not going to the playoffs next year.

I actually think Lugo and Drew represent a step up defensively from their predecessors, Pedroia I haven't seen play enough, tho he might well be better with the glove than Loretta.

As for the projections:

I'm not sure what possible reason there is to think that Varitek, a rapidly aging catcher, will bounce back this season. If anything, he could be even worse than he was last season, when his bat had the punch of an old lady - with arthritis.

Pedroia's excellent minor league career gives reason for hope that he could produce the type of rookie season envisioned, but will he actually do it?

Lowell had an excellent start last season, iirc, and a miserable finish. I suspect his entire year in 07 will look more like his 2nd half.

I expect Papi to be his typical brilliant self, but I'm not so sure Manny won't fall off a bit, to an ops in the high .900s, which is still excellent.

Lugo's a significant improvement over AGon. And Drew likewise ought to be over Trot, but I'd be surprised if this translates into the Sawx scoring over 950 runs.

As for the pitching: it's improved to, but for all the talk of Bye-Bye Beckett's peripherals, his era was so high not b/c he was unlucky, but b/c teams slugged the heck out of the ball against him. I suspect he'll be somewhat better, but not the top of the rotation guy the Sawx thought they'd get.

I also think Schilling is getting old, and fat, fast, and will likely drop off significantly, tho he may start solid. Can Wake hold up? Can Mats make the transition to MLB? Can Paps make the transition from the pen, where hitters only had the once chance to hit him in a game, rather than 3 or 4? These are all question marks.

You can't just round 1.83 up to 2 -- as he says he did -- when using the pythagorean therum. Had he done the calculation correctly, the result would be 93 wins, which is much more realistic.

No, the result with a 1.83 exponent would be 103 wins -- not much difference.

These projections are only slightly more realistic than me taking my latest Angel season results off my PSP (123-37 with 2 games to play).

He's projecting runs based on how many runs per game the starting 9 give you, and runs allowed based on what the 11 possible opening day pitching staff would allow, projected to a full season of innings.

I can guarantee this: The Red Sox (and every other team) will give at bats to a lot more than 9 batters, and innings to a lot more than 11 pitchers.

And the bench/ injury replacements/ minor league callups will be a lot worse than the starters.

If you applied this kind of analysis to even the Royals you'd probably get close to an 81 win team.

Rally is exactly right. You have to figure that something like 25% of the PAs will go to other players, and perhaps a higher percentage of the IP to lesser pitchers. You probably need to reduce RS and increase RA by 5-6%. That would take you from 103 wins to something more like 90 (just a ballpark estimate).

Similarly, you can't use the lineup tool to estimate how much the Red Sox offense underperformed in 2006 in terms of RS vs. hitting stats. It was more like 22 runs (looking at BPro's "EQA runs") than 83.

In fact, if you look at the PECOTA win totals he cites for the starters, and RBI for hitters, you'll see that PECOTA itself projects much less playing time.