Thursday, September 04, 2003
D'oh!
I spend far too much time with the normal distribution...
As Tom points, out, we are actually dealing with a binomial distribution, and the variance for that is np(1-p) = 162(.5)(.5) = 40.5. Take the square root of that to get about 6, so 87 wins would be a standard deviation above .500. So in every year, we have had a team a standard deviation above...
Of course, it is a little more complicated than this, because you have 16 (or 14) teams shooting for the playoffs, so simply looking at a standard deviation and comparing it to the actual results is probably too simple a model. If I have the time and motivation, I'll whip up a simulation and run it a few tens of thousands of times.
posted by David 7:14 PM
