In this model, we consider the mean points-per-game achieved by each team. To get this, you simply add up all the points they socred to date in all of the games in which a given team played, and then divide by the number of games played by that team. The repeat for all teams in the league. To make a prediction, just look at which team has the most points. The 2014 NFL season performance of this algorithm is shown below:
Considered over the 2012,2013 and 2014 NFL seasons, this model makes the corect call 61% of the time.
Similarly to the wins Percentage model, the mean points model above does not account for the home-team advantage. In this case an adjustment can be made by just adding 3 points (the mean home-team advantage) to the mean score when a team is playing a home game. Doing so improves the 3-year correct-call average from 61% to 64%, which is a small but noticable improvement.
Professional odds-makers have better predictive methods and algorithms than this one. This algorithm will NOT let you beat the odds in a consistent manner in Las Vegas-style gambling.
Next Page - Margin-of-Victory Best Fit