2024 NFL Model
Last Updated December 12 2024.
To date, our model has correctly picked 70.2% of games, the eighth most accurate of models tracked here.
This page contains our 2024 NFL model including power rankings and simulated championship odds, and model accuracy and downloadable predictions for the rest of the season! Also at the bottom of this page is an FAQ. Our methodology is as follows:
- Create rankings using Vegas data. This results in up to date rankings that reflect current injuries, outlooks, and performance to date. To read more (especially about how accurate this method is compared to other models), we have a full write up about Implied Vegas Ratings. To read why Vegas ratings make sense read about Ensemble Learning in Sports and the Efficient Market Hypothesis in Sports.
- Estimate super bowl odds using computed team ratings, current records, and 100,000 simulations of the remainder of the season.
- Evaluate model on past games and predict the spread of the remaining games this year
2024 NFL Power Rankings
The graphic below shows the current NFL team power rankings and how teams have moved week-to-week. The numbers in this plot are rankings; to see exact team ratings for more granular comparisons, use the table after the graphic.
The ratings in this table represent “points better than an average team”. The difference between two teams is a prediction of their spread on a neutral field. Adding two points to either team accounts for home field advantage
| Team | Rating |
|---|---|
| BAL | 7.4 |
| DET | 6.7 |
| BUF | 5.7 |
| PHI | 5.1 |
| GB | 4.8 |
| KC | 4.3 |
| MIN | 3.4 |
| WAS | 2.5 |
| LAC | 2.2 |
| MIA | 2.1 |
| HOU | 1.9 |
| CIN | 1.9 |
| PIT | 1.8 |
| ARI | 1.5 |
| DEN | 1.3 |
| SF | 0.7 |
| SEA | 0.7 |
| TB | 0.1 |
| LAR | -0.2 |
| IND | -0.3 |
| ATL | -0.3 |
| NYJ | -1.2 |
| CHI | -1.5 |
| CLE | -2.5 |
| NE | -3.6 |
| TEN | -3.9 |
| LV | -5.5 |
| NO | -5.8 |
| JAX | -5.8 |
| CAR | -6.4 |
| DAL | -6.8 |
| NYG | -10.2 |
2024 Super Bowl Odds
The following graphic shows how the super bowl odds have evolved week-by-week. Only teams that have eclipsed 5% in any individual week are given a dedicated slot, the other teams are grouped into ‘other’. This graphic is meant to be a visual representation, the table after gives exact numbers.
The table below shows our model’s estimated super bowl odds. Also contained is probabilities of making the super bowl, making the conference championship game, making the divisional round, and making the playoffs. These all resulted from 100,000 simulated seasons using our Implied Vegas Ratings.
| Team | SB% | Win Conf% | Win Div% | Win WC% | Playoff% |
|---|---|---|---|---|---|
| DET | 23.9 | 42.4 | 62.5 | 93.9 | 100.0 |
| BAL | 20.7 | 35.6 | 54.4 | 73.0 | 99.0 |
| KC | 9.6 | 18.6 | 31.9 | 66.0 | 100.0 |
| BUF | 7.9 | 14.5 | 21.8 | 67.9 | 100.0 |
| PHI | 7.4 | 14.0 | 20.4 | 64.1 | 100.0 |
| PIT | 5.9 | 14.3 | 30.4 | 66.0 | 99.7 |
| MIN | 5.6 | 14.4 | 34.7 | 60.2 | 100.0 |
| GB | 5.2 | 10.6 | 25.4 | 53.2 | 97.6 |
| HOU | 3.2 | 6.5 | 26.3 | 53.0 | 87.5 |
| SEA | 2.6 | 7.8 | 16.2 | 31.4 | 67.6 |
| LAC | 2.0 | 5.6 | 18.8 | 34.0 | 92.1 |
| WAS | 1.9 | 3.7 | 9.5 | 32.7 | 85.4 |
| TB | 1.5 | 4.7 | 17.7 | 36.4 | 79.2 |
| DEN | 1.4 | 3.3 | 11.7 | 25.9 | 79.9 |
| LAR | 0.4 | 0.9 | 4.8 | 11.0 | 27.8 |
| ARI | 0.2 | 0.6 | 2.3 | 5.5 | 11.9 |
| SF | 0.2 | 0.4 | 1.2 | 2.0 | 4.8 |
| IND | 0.2 | 1.1 | 3.8 | 10.7 | 29.8 |
| ATL | 0.2 | 0.5 | 5.2 | 9.3 | 24.9 |
| MIA | 0.0 | 0.4 | 0.8 | 2.8 | 8.9 |
| LV | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 |
| DAL | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 |
| CAR | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 |
| JAX | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 |
| NO | 0.0 | 0.0 | 0.1 | 0.3 | 0.8 |
| NYJ | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 |
| TEN | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 |
| NE | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 |
| CLE | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 |
| CHI | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 |
| CIN | 0.0 | 0.1 | 0.1 | 0.7 | 3.1 |
| NYG | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 |
Accuracy and Predictions
Year to date, our model has accurately predicted 75% of games. The figure below shows how our accuracy has changed week-by-week.
Our predictions for the rest of the season can be downloaded here.
FAQ for 2024 NFL Model
How Do Implied Vegas Ratings Work?
If Kansas City was favored at home by 3.5 against Baltimore week 1 and 5.5 at home against Cincinnati week 2, this means that Baltimore is about 2 points better than Cincinnati. By combining spreads over multiple weeks, our model figures out the implied quality of every NFL team by finding similar “second order comparisons”.
More precisely, our model solves a least squares regression problem to find each team’s rating that best explains the Vegas spreads we’ve observed. In doing so, it also corrects for home field affects worth 2 points. Because team quality can change rapidly due to injuries or realizing teams aren’t very good, our model solves a least squares regression with a rapidly fading memory, only using the previous 3-4 weeks of spreads.
How Accurate are Implied Vegas Ratings?
Though we do not track week-by-week results currently, our longer study posted here shows that Implied Vegas ratings, and therefore our 2024 NFL model, are as accurate as the Vegas line is. By looking at prediction tracking here, this means that our 2024 NFL model is among the most accurate computer rankings available online.