Sunday, September 25, 2016

Pythagorean Win Percentage Calculated from Spread and Over/Under

   I'm sure somebody has made this calculation before, but I couldn't find it on the web, so here is my version of calculating the predicted win percentage of a football game, based on the spread and over/under.   You get the spread and over/under from or most gambling sites and sports sites.  

   The idea of Pythagorean win percentage is that if you know how many points a team gives up and how many they score, you can estimate the win percentage.   as follows:

Win pct = (points scored)^2/ [(points scored)^2 + (points allowed)^2]

OK, so the football betting line gives you the Las Vegas estimate (about the best in the world) of  point differential = points scored-points allowed.

And the overunder is the best estimate of 
total points = points scored plus points allowed. 

So,  you can calculate an estimate for points scored =  1/2 *   (overunder  plus spread) . 

An estimate for points allowed is 1/2 (overunder - spread).  

As an example, on Sunday Sept 25, the Miami Dolphins are favered by 7.5  points, with the overunder at 42.5  

Predicted Miami points = 0.5*  (42.5 + 7.5) = 25

Predicted Cleveland points =0.5 * (42.5- 7.5)  = 17.5

Miami estimated win probability =  (25^2)/(25^2 + 17.5^2) = .671

If you could play like thatl all season, that's pretty good.  For example an 11-5 team is .688, 10-6 is  .625.  

By the way, I calculated this just for fun.  It's not intended to be a betting tool.  I'm not a bettor, because I believe that 10% fee takes pretty good care of the bookies, and they are better than us.   

No comments:

Post a Comment