miscellany

basketball and football

first, as of last night, every nba team has now played at least 20 games this season.  here are your top ten (according to my computer):

miami, boston, dallas, san antonio, la lakers, new orleans, utah, orlando, chicago, denver.

second, i’ve been trying my hand at the nfl as well.  this has proved considerably more difficult than i anticipated for two main reasons.  (1) the season is very short, so there are fewer statistics to work with.  (2) lots of games are decided by statistical anomalies (e.g. when (i) michael vick throws an interception and (ii) that interception results in a 14-point swing).

i originally developed two different models, which i thought, a priori, would yield very similar results.  here are your top 10 according to each:

method 1: philadelphia, pittsburgh, new england, green bay, kansas city, san diego, tennessee, ny jets, chicago, ny giants.

method 2: san diego, ny giants, green bay, kansas city, tennessee, philadelphia, new england, atlanta, pittbsurgh, houston.

as you can see, even though these two lists contain a similar set of names, the ordering is very different.  and there are two other striking issues.  (1) philadelphia is hard to make sense of statistically because they are one of the best teams in the league (according to any measure i can devise) with michael vick and they are below average without him.  given the small data set, it’s impossible to know what to do about that.  (2) san diego doesn’t win very much.  according to every model i can come up with, they should win a lot.  it’s hard to know whether my model is wrong or san diego is unlucky, and the shortness of the nfl season provides no help here at all.  in any case, my two original methods can be easily combined to yield something that seems intuitively better than either of the two above lists:

method 3: philadelphia, green bay, new england, pittsburgh, san diego, kansas city, ny giants, tennessee, chicago, houston.

basketball

every team in the nba has now played at least ten games this season, which is enough to have some meaningful statistical results.

over the past couple of years, i’ve developed three different (fairly reliable) statistical methods for rating the performance of different teams.  here are the top ten teams in the nba (in descending order) according to each of them.

method 1:

miami, orlando, la lakers, new orleans, boston, san antonio, dallas, atlanta, indiana, chicago

method 2:

miami, new orleans, dallas, la lakers, boston, orlando, chicago, san antonio, milwaukee, denver

method 3:

miami, new orleans, dallas, la lakers, boston, orlando, san antonio, chicago, denver, utah

in the past, method 3 has been the best predictor of future success and method 1 has been the worst.

my sunstone paper

this is the version of my paper that i presented at sunstone today.  it is not really publication ready, but i had a good time presenting it.

question for philosophers or others

if i tell you to “do it for you, not for me” i am telling you to do something that you can’t do because i told you to.  if i hand you the keys to my new ferrari and say “take it for a spin” i am telling you to do one thing (drive my car) in hopes that you will come to do a different thing that you couldnt have done just because i told you to (drive my car just for the sake of driving it).  can anyone think of a clear and common case of ordinary english (that doesn’t involve manipulation or something of the sort) in which i do both (i.e. tell you to do something that you can’t do because i told you to in hopes that once you do that thing you will come to do something else that you couldn’t have done because i told you to)?