how many games a reasonably healthy 2010 Sox would have won (I think about 94 games),

Just a thought but I think this is light by 2-3 games. Again, we're talking about degrees but through the series in San Francisco the Sox were on pace for 97 wins. Obviously that's a selective end point but it is the spot where Pedroia, Martinez and Buchholz all went down and for the rest of the year the Sox were missing at least two of Pedroia, Youk and Martinez, ostensibly the middle of their lineup.

Moreover, their 2nd order wins on BPro had them winning more games than they actually won. This has some relevance for what we think about them for next year.

The Red Sox were functionally eliminated pretty early in September too, right? While I can't remember a lot of frustrating "Mark Loretta as starting first baseman" lack of emphasis on winning this past September, it's got to be some kind of factor.

If the point of the projection is to see if they make it to 95 wins, and are thus in contention all season, I'd add at least 1 win to those 89 to reflect a team that's pushing hard at the end, e.g. higher pitch counts for still-effective starters, multiple innings for a possibly sore-armed Bard, starters playing every day even when banged up, etc.

I'd never add or subtract wins from last year's projections. If you do that for all teams, I think you'll find that the league wide win % will be something like 52 or 53%, which, of course, is impossible.

To project wins, you take a team's individual parts, project performance, project playing time, and sim. Last year's performance only factors in so much as it effect's next year's individual projection.

To project wins, you take a team's individual parts, project performance, project playing time, and sim. Last year's performance only factors in so much as it effect's next year's individual projection.

Yes, that's how you do a "proper" projection, but that isn't what's being discussed here. MCoA is trying to make a q&d projection based on adding to and subtracting from last year's win total (and has acknowledged this isn't a real projection).

And since no one is talking about simulating the league, I'm not sure (really, I could be convinced) whether it matters that the league's winning percentage would be over 50% with a "full effort" bonus. We're only comparing the quality of the '10 Red Sox to the '11 Red Sox using '10 win total as a measuring stick, and I'm pointing out that since the end of the stick is slightly floppy it's therefore a little inaccurate.

Oh, OK. I completely agree that you don't compare a team to actual wins. Those wins contain luck, both good and bad, and are very different sets of players. Even the same team with the same players will perform differently. The 2011 team with the same personal shouldn't be expected to have a 700 OPS at CF and LF. To do q&d, I typically have a feel for a team that was previously projected at a certain level (e.g., 2010 RS pre-season = 94 wins) and then adjust based on that and expected parity level.

At this level, it is so much guestimating that if someone tries to convince you that there is a difference b/w 93 and 95 wins they see way too much precision.