The Sims

by
April 7, 2009

If you’re a reader of this blog this kind of thing is not news to you, but it’s interesting all the same:

“Computer simulations work pretty well in baseball for two reasons,” said Carl Morris, a professor of statistics at Harvard University who has written several papers that commingled baseball and formal statistical theory. “In general, they allow you to study fairly complicated processes that you can’t really get at with pure mathematics. But also, sports are great for simulations — you can play 10,000 seasons overnight.”

No one can afford to wait less than major league teams, which crave every extra run or victory they can wring from their $100 million rosters. John Abbamondi, the assistant general manager for the St. Louis Cardinals, says his team and about 10 others use simulations to evaluate potential trades and how they might affect the pennant race.

“It’s all part of the statistical analysis that complements the more traditional scouting we do,” he said.

Not that they can’t be misleading. For example, back in 1988 I loaded up my Lance Hafner Baseball team (for Commodore 64) with guys like Pedro Guerrero’s 1978 season (.625/.625/.875 in 8 PAs) and Cesar Cedeno’s 1985 (.434/.463/.750 in 82 PAa). The results were fantastic — I went 162-0 vs. the 1987 Major League Season, just edging the Blue Jays by 63 games — but I can’t help but think that they my strategic skills were developed all that much. Especially considering that I went in and zeroed out everyone’s errors before the season. Everyone on my team anyway.


3 Comments
Oldest
Newest Most Voted
Inline Feedbacks
View all comments
Chris H.
15 years ago

I used to play SSI’s Computer Baseball on the Apple II, way back when.  It was stats-based instead of being an arcade-ish game; you were in charge of strategy, not swinging the bat.  You could set things like running strategy (conservative, normal, or aggressive), infield depth, and so on.

You could also put in the stats for your favorite team and play them against whoever you liked.  (It only came with a smattering of “all-time great” teams, but of course you could type in others.)  The game would calculate ability based on the stats, but it really didn’t understand sample size.  It also didn’t keep track of stats based on game results, so after every game you had to manually update the stats.

Of course, since it didn’t understand sample sizes, one or two lucky games could profoundly affect long-term performance.  You have no idea how many batting titles Thad Bosley won on my Cubs team.  (He hit .418 one year!)

I’m sorry; what were we talking about again?

0
Pete Toms
15 years ago

Somebody at THT interviewed Pete Palmer (last year, IIRC) and he admitted to not watching (or rarely watching, I can’t recall) baseball.  He just loves the math.  This is an extreme example of why managers and scouts can be contemptuous (sic?) of objective analysis types.  Not only did they not play the game (or at least at any highly skilled level), some might not even enjoy watching it.  I can understand why it sticks in their craw (see LaRussa’s comment in the piece) but this debate is over, the stat heads are just WAY smarter.

The Numbers Game is a very enjoyable read about the evolution of objective analysis, much better than Moneyball.

When will this take root in football?  There was a paper making the rounds on the internet a year or two ago explaining why coaches should punt less and “go for it” on 4th down.  I recall somebody questioning if the football coaches were smart enough to understand the math however.

0
Chris H.
15 years ago

Footballoutsiders.com has dome some good work with respect to objective analysis, but it’s clearly not nearly as mature as baseball analysis is.

0