Friday, January 18, 2013

2013-01-18: NFL Conference Championship Predictions

The NFL Conference Championship games are this weekend and just one game separates the four remaining teams from Super Bowl XLVII. If you ignore the vapidity of the Te'o coverage, there is much discussion of how the loss of Rob Gronkowski will impact the performance of the Patriots this weekend. Aaron Hernandez should perform admirably and the entire team should be able to make up the difference and triumph.

For our predictions we run a number of different types of algorithms in our research and compare the outputs. The three main algorithms that have consistently had the best performance are a Support Vector Model (SVM), a Multilayer Perceptron Neural Network, and a ranking algorithm. All three algorithms sided with the favorites.

The SVM gives us a binary output, winner or loser. There is nothing in between. The SVM chose San Francisco and New England.

The Neural Network output is a continuous variable that is supposed to be the margin of victory. A positive score favors the home team and a negative score favors the visitor. The San Francisco-Atlanta game had an output of -0.98 so San Francisco is a one point favorite, this will be a close game. With the spread currently at three to four in the 49'ers favor, Atlanta may cover the spread. The New England - Ravens game had an output of 8.47, With the spread currently at about nine points the Ravens will probably cover.

The ranking algorithm is based on Google's PageRank algorithm which we have outlined in previous posts. San Francisco and New England are ranked higher than their opponents so they predicted to be the winners . The rankings for the entire 2012 season are visually represented the the figure below.


Combing the algorithms together with a majority voting scheme results in the following teams being the predicted winners straight up.

New England
San Francisco

No matter how you look at it, this weekends games should be very good.

--Greg Szalkowski

No comments:

Post a Comment