Season projections
Loading…
| # |
|---|
Elo is each team's rating going into its next game, including the adjustment for that game's starting quarterback (hover for the breakdown). Tiers by Elo: . Win conf. means winning the AFC or NFC title game and reaching the Super Bowl. Wk Δ is the change in playoff odds from the week before, in percentage points. Select a team to see how its odds have moved.
Game forecasts
Order
Game quality (0 to 100) blends how close the game should be (45%), how good both teams are (47%) and both starting quarterbacks (8%). Team and QB scores use a harmonic mean, so one great team facing a bad one doesn't make a great game.
Leverage comes from the season simulations: every team's odds in seasons where the home team wins are compared with seasons where the away team wins, and the probability that changes hands across all 32 teams is added up, with simulation noise stripped out. Playoff odds count 60%, division titles 20%, #1 seeds 10% and Super Bowl odds 10%, each scaled by how many there are (14 playoff spots, 8 division titles, 2 #1 seeds, 1 Super Bowl), so the result reads in playoff-odds points.
Elo is each team's rating going into the week. Rest adjusts for days since the last game (short week or extra rest), Field is home-field advantage (smaller at neutral sites), QB adjusts for the starting quarterback, and Travel for distance traveled. Game Elo adds them up, and the gap between the two Game Elos sets the predicted margin and win chance. GQI (game quality index, 0–100) rates how strong the matchup is. The tier under each team uses its Elo with the QB adjustment: . Past weeks show each game as forecast before kickoff, with the final score and whether the favorite won.
What-if
Pick the winner of any game, past or future, then simulate the rest of the season again.
Results
Pick some winners, then run the simulations.
| Team | Playoffs | Win div. | #1 seed | Win SB | Proj. wins |
|---|
Each run locks in your picks and simulates the rest of the season with the same model as the main forecast. The arrows compare it with the same number of seasons and the same random draws without your picks, so they show the effect of the picks rather than noise. Picking a future winner also counts as evidence about how good the two teams are: seasons where that result was more likely count for more, the same way the rooting guide works, so the more games you pick, the fewer seasons effectively count. Changing a past result changes the standings, not the Elo ratings that came after it.
Rooting guide
| Game | Root for | If they win | If they lose | Swing | Kickoff |
|---|
Root for is the result that helps the selected team more. The arrows show how far that team's odds move from where they are now, in percentage points, and Swing is the gap between the two results. Only games not yet played are listed.
Win or lose
Each team's odds now, with a win, and with a loss in its game this week. Teams on a bye or already finished show one dot.
With a lossNowWith a win
Simulate the season
One way the rest of the season could play out, from the current ratings. Every click plays a new one.
Playoffs
Higher seed at home · the No. 1 seed hosts the lowest seed left after the Wild Card round · Super Bowl at a neutral site
Final standings
Playoff seed on the left · point differential includes games already played
Week by week
Each game's winner follows the model's odds, using the same ratings, home field, rest, travel, QB adjustments and season-long uncertainty in team strength as the forecast. Scores are built drive by drive: about 11 drives a side, each ending in a touchdown (usually 7, sometimes a missed extra point or a 2-point try), a field goal or nothing, plus the odd safety. How many points a team scores and allows comes from splitting its rating into offense and defense using the points it has scored and allowed this season (pulled toward an even split early on) and its starting QB. Tied games go to overtime. This is one random season; the Projections tab shows the averages over thousands of them.
Quarterback Elo
Draft projection
| Team |
|---|
Pick distribution
Share of simulated seasons the pick lands in each slot. Dark bars are top 5, medium bars 6–10.
Projected first round
Every first-round pick, ordered by average draft slot.