Future Win Totals Thinking, Take 3

On Walter's suggestion, I played with the binomial distribution. It's not perfect, but it's better. I now get an expectation of 1.53 teams going undefeated and 2008 USC with a 33.85% chance of having exactly one loss and 19.02% chance of going undefeated, both of which make a lot more empirical sense.

This isn't perfect, because I tried to reverse engineer the E(wins) out of the binomial distribution, until I realized I couldn't get half wins back out, so the E(wins) calculation is using the same assumptions as before (e.g. symmetry).

In any case, I'm not really planning on using this for anything other than making fun of retard fanboys who think Georgia is going to win the national title (these numbers give 2008 UGa a 17.1% chance of exactly one loss and a 5.1% chance of going undefeated -- prescient), so it doesn't need to be perfect, but I think Jonny will be able to write some good posts with these.


Walter said...

Yay, I helped!