Friday, January 30, 2015

Technical T-Rank Note

Close observer(s) of the T-Rank may have noticed that all the top teams' Barthags dropped last Friday. Very astute observer(s) would have noticed that the worst teams' Barthags went up.

This happened because I lowered the "exponent" used to calculate the Barthag. This is a constant in the formula used to calculate the Pythagorean Expectancy. Changing it doesn't change the rank order of teams, and different Barthags don't change how the projected scores of games are calculated -- but it does change the projected certainty of the result. In other words, that percentage figure in the T-Rank predictions. That figure is actually important, because it is what I use to project future records. (For example, a team that is projected to have a 50% of winning game 1, an 80% chance of winning game 2, and a 70% chance of winning game 3 will have a projected record of 2-1 because .5 + .8 + .7 = 2.)

I lowered the T-Rank exponent from 10.25 to 9.0 because favorites were not winning as often as projected. Also, the projected records for the top teams just seemed overly optimistic to me.

So that's the explanation for that.

No comments:

Post a Comment