NFL 2015 Week 10 Predictions – Zoltar is Crushing Bing

Every year I write a computer program that predicts the results of NFL football games. It’s a fascinating problem and one where you know quite quickly how good the predictions are. The prediction system is named Zoltar, after the fortune telling machine you see in arcades.

Zoltar is designed to predict the margin of victory (or loss) for each team. This is a much harder problem than just predicting the winner.

Microsoft Bing predicts NFL winners each week and publishes results, but only when they do well. I was curious about how Zoltar is doing this season compared to Bing. Through the first 9 weeks, Bing has an accuracy of 89 / 132 = 67% and Zoltar has 102 / 132 = 77% so Zoltar is doing much better than Bing.


Zoltar’s point spread predictions for week 10 of the 2015 NFL season are:

Zoltar favorite =        jets  by    0  underdog =       bills
Zoltar favorite =     packers  by   10  underdog =       lions
Zoltar favorite =    panthers  by   12  underdog =      titans 
Zoltar favorite =      eagles  by    6  underdog =    dolphins
Zoltar favorite =    steelers  by   11  underdog =      browns 
Zoltar favorite =        rams  by    6  underdog =       bears 
Zoltar favorite =      ravens  by    8  underdog =     jaguars
Zoltar favorite =  buccaneers  by    0  underdog =     cowboys
Zoltar favorite =    redskins  by    0  underdog =      saints 
Zoltar favorite =     vikings  by    2  underdog =     raiders 
Zoltar favorite =     broncos  by   10  underdog =      chiefs
Zoltar favorite =    patriots  by    5  underdog =      giants 
Zoltar favorite =    seahawks  by    0  underdog =   cardinals 
Zoltar favorite =     bengals  by   11  underdog =      texans 

Notice several of the predictions are a tie game. When this happens, in order to calculate a head-to-head winning percentage, Zoltar predicts the home team will win.

Zoltar’s predictions agree closely with the Las Vegas betting line. The four games with a significant difference are:

* Zoltar says the Panthers will win by 12 over the Titans but Vegas says Panthers will win by only 5.5 so Zoltar recommends betting on the Vegas favorite Panthers.

* Zoltar says the Steelers will win by 11 over the Browns but Vegas says the Steelers will win by only 4.5, so Zoltar recommends betting on the Vegas favorite Steelers.

* Zoltar says the Vikings will win by 2 over the Raiders but Vegas says the Raiders will win by 3, so Zoltar recommends betting on the Vegas underdog Vikings.

* Zoltar says the Broncos will win by 10 over the Chiefs but Vegas says the Broncos will win by only 6.5, so Zoltar recommends betting on the Vegas favorite Broncos.

Zoltar does not take into account key injuries for the Steelers-Browns and Vikings-Raiders games.

Update (Tuesday morning 11-17-2015): Zoltar went 3 for 4 against the Vegas spread in week 10.

Panthers beat the Titans 27-10 (correct prediction)
Steelers beat the Browns 30-9 (correct prediction)
Vikings beat the Raiders 30-14 (correct prediction)
Chiefs beat the Broncos 29-13 (incorrect prediction)

In head-to-head predictions (meaning just predicting the winner of a game, not the point spread), Zoltar went a very poor 6-8. Bing did even worse, going a horrendous 4-10.

This entry was posted in Machine Learning. Bookmark the permalink.