I hate how people are trying to apply statistics to basketball, like it’s some voodoo that could potentially unlock all sorts of answers and reduce the sport to a science. It’s a ridiculous, depressing and self-defeating (for sports fans, anyway) thing to do. The pompous proponents of this mentality are correct in arguing that there is value to applying predictive analysis techniques to the reams of numbers basketball statisticians have been keeping for years, but the purportedly sophisticated methods that people like John Hollinger are pushing in order to allow, for example, comparisons between players of different eras, are so desperately arbitrary and leave so much information out that you really wonder what they’re smoking (and yes, wish they’d pass it around).
All I want right now, is for someone to look at Tim Thomas’ line in the March 28 Bulls-Pacers game and restate the value of +/-, basketball’s statistic of the moment:
MIN FGM-A 3PM-A FTM-A OREB DREB REB AST STL BLK TO PF +/- PTS
Tim Thomas, PF 6 0-3 0-1 0-0 0 1 1 0 0 0 0 2 +8 0
Yes, for the 6 minutes in which he netted 1 rebound, 3 misses and 2 fouls, he was rewarded with a +8 rating.
I was never able to figure out the /- rating system. The bulls are pretty much screwed no matter what spot, 7 or 8, they end up in. Actually I think the bulls are better off playing against the Celtics, at least they can lose to the champs, plus if Garnett is out they might actually win a game or two. If they get to the 2nd round of the playoffs I’m gonna splurge $300 and get the NBA channel.
Another example of how misleading these stats can be: http://www.sonicscentral.com/apbrmetrics/viewtopic.php?t=2166&sid=7204824dc5a5d0b3ab4a3be8033fc02f
As for your question about whether this same argument holds for the use of advanced statistical techniques in pricing derivatives…no. Derivative pricing theory is based entirely on replication arguments; that is, given that some instruments in the market trade at certain prices, what price would some other instrument have to trade at in order to be consistent. This is also why it’s practically impossible to naively make money with derivatives — you have to venture away from risk-neutral pricing, making assumptions and approximations that are correct on average.