clock menu more-arrow no yes

Filed under:

sim mania

New, 117 comments

tangotiger is running a very cool experiment on clutch hitting; i urge you all to participate.

the Replacement Level Yankees Weblog’s annual Diamond Mind Blowout has become one of my favorite spring traditions; 2008 marks the 4th year RLWY has crunched the numbers in this fashion. if you’re not familiar w/ the Blowout, RLWY takes various sets of projection data for the upcoming season --- PECOTA, ZIPS, bill james, etc etc --- plugs them into the Diamond Mind simulation game, and runs 1,000 full-season simulations for each set of data. then it rolls the results from all the different projections into a single database comprising several thousand simulated seasons. all the usual disclaimers apply --- projections are not 100 percent accurate, the game is played by people not computers, the projections all had st louis losing to detroit in the ’06 world series, blah blah blah. all acknowledged up front. they’re only projections; don’t take ’em too seriously.

the Blowout had a very good year in 2007. take the nl central --- here are RLYW’s predicted w-l records, vs the actual results:

cubs 84 85
mil 81 83
stl 84 78
hou 80 73
cin 73 72
pitt 73 68

didn’t miss the mark by much on any team. indeed, RLYW’s projected win total in 2007 was within 3 games of the actual total for half of the big-league clubs; it was off by 10 games or more for only 3 (seattle, colorado, and minnesota). full results of last year’s projection are at this link; you have to scroll down about halfway to get the final, all-in-one projection.

ok, enough preliminaries. for the Blowout this year, RLYW ran simulations for six different sets of data --- PECOTA, ZIPS, CHONE, Diamond Mind, Hardball Times, and CAIRO (which is RLYW’s own, brand-new projection system). they played the 2008 season 6,000 times --- so many times that it took 2 separate posts to present all the results (part 1 here, part 2 here). in the final, aggregate projection (see part 2), the cardinals won the division a little less than 10 percent of the time --- 503 times, to be exact. sounds about right to me. they also won 130 wild cards, so they made the playoffs about 11 percent of the time. their mean average record? 78-84 --- exactly what the VEB community projected a couple weeks back. indeed, we as a community are almost in lockstep w/ RLYW for the entire nl central:

cubs 88 88
mil 85 88
stl 78 78
cin 77 81
hou 75 75
pitt 70 70

4 out of 6, right on the nose; the wisdom of crowds strikes again. over the 6,000 sims, the cards averaged 762 runs per season (10th best in the national league) and 784 runs allowed (9th best). the st louis pitching staff comes off as being not much worse than chicago’s (31-run difference) or milwaukee’s (9 runs). the thing that separates the cards from the division leaders is their much weaker offense --- they’re 60 runs worse than either chicago or milwaukee per this exercise.

here are the cards’ average won-loss records, broken out by individual projection system (1,000 sims for each system):

83 78 73 76 80 80

PECOTA and Diamond Mind see things quite a bit differently, no? in the entire exercise, there were only 4 cases in which two projection systems differed by 10 games or more on a given team.

alright, so much for that. a few other random notes: