Find the 2017 season here HERE.
UPDATE: I’ve moved the 2016 AMOS model to a more permanent page, along with a revamp of the dashboard. Please keep up with the season HERE.
For those just joining, Amos is a statistical model created to predict the outcomes of each NFL game. Amos takes into account 224 different data points to compute three different probabilities for each game.
First, Amos calculates the probability of each team winning. The dashboard below then displays the team which has the greatest probability of winning. Second, given the spread that has been assigned to a particular game, Amos calculates the probability of that team covering the given spread. Finally, given the Over/Under assigned to a particular game, Amos predicts the probability of both teams’ combined scores to break that threshold.
Additionally, Amos then forecasts the remainder of the season and calculates the most probable ending record for each team.
Amos also has peers within the NFL prediction field. While there are a number of sources for game predictions, I have identified Microsoft’s Bing Predicts (Bing), ESPN’s Football Power Index (FPI) and Nate Silver’s Elo (Elo) as benchmarks for comparison to Amos this year. The selection is due to the statistical approach of each of these predictions methods, which provides similar, but not exact, grounds for comparison.
As a summarization, the below chart plots all four models’ confidence by game. Dots plotted below 50% represent an away favorite and dots plotted above 50% represent a home favorite.
The farther a dot is away from the center, 50%, the more confident a given model is of its predicted favorite.
Based on historical data and current team data, Amos also computes probabilities of over/under and spreads being covered.
Given the over/under assigned to a matchup, Amos computes the probability that both teams’ combined score will push or cover the over/under. For example, an over/under of 41 has been assigned to a game and Amos has computed a probability of 56%. This means that if the same match up could be played an infinite amount of times, 56% of the time the teams’ combined score will be equal to or greater than 41. To be able to evaluate Amos, if Amos’ assigned probability is above 50% and the teams’ combined score is equal to or greater than the assigned over/under, Amos will be given credit for being ‘correct’. The same is true if Amos’ assigned probability is below 50% and the teams’ combined score is less than or equal to the assigned over/under.
Spreads are slightly more complex. For consistency within my modeling techniques, a negative spread represents a home team favorite and a positive spread represents an away team favorite. Given this, Amos computes the probability that the favored team will cover the spread assigned to the game. For example, a spread of 3 has been assigned to a game and Amos has computed a probability of 39%. This means that if the match could be played an infinite amount of times, only 39% of the time the away team would win by 3 or more points. If 3 is swapped out with a -3 in our previous example, then the interpretation would change to only 39% of the time the home team would win by 3 or more points.
Have thoughts on the predictions? See a missing game or an incorrect score? Leave a comment below or send us an email at TrevorBischoff@gmail.com
See Amos’ predictions from the 2015 NFL season:
Week 1: 11-5
Week 2: 9-7
Week 3: 11-5
Week 4: 9-6
Week 5: 10-4
Week 6: 10-4
Week 7: 9-5
Week 8: 10-4
Week 9: 7-6
Week 11: 10-4
Week 13: 11-5
Week 14: 9-7
Week 15: 14-2
Week 16: 7-9
Week 17: 9-7