NFL 2026 Season Projections
Projected win totals for all 32 teams from 5,000 simulated seasons, a win probability for every one of the 272 scheduled games, and a head-to-head tool — from an Elo model trained on 6,967 real games (1999–2025). Every projection is shown as a range, because that’s what preseason knowledge honestly looks like.
Head-to-head matchup probability
Pick any two teams — probability uses each team’s entering-2026 rating plus the standard 48-point home edge.
Probabilities, not predictions — a 65% favorite still loses about one time in three.
Projected win totals, 2026
Sorted by projected wins. The range is the middle 80% of that team’s 5,000 simulated seasons (10th–90th percentile).
| Team | Entering Elo | Proj. wins | 80% range | ||
|---|---|---|---|---|---|
| 1 | Seattle Seahawks | 1635 | 11.3 | 8–15 | |
| 2 | Denver Broncos | 1614 | 10.7 | 7–14 | |
| 3 | Houston Texans | 1599 | 10.6 | 7–14 | |
| 4 | Los Angeles Rams | 1598 | 10.4 | 7–14 | |
| 5 | Buffalo Bills | 1596 | 10.2 | 6–14 | |
| 6 | Jacksonville Jaguars | 1576 | 10.1 | 6–14 | |
| 7 | Detroit Lions | 1558 | 10.1 | 6–14 | |
| 8 | Philadelphia Eagles | 1557 | 9.9 | 6–14 | |
| 9 | New England Patriots | 1569 | 9.6 | 6–13 | |
| 10 | Minnesota Vikings | 1556 | 9.6 | 6–13 | |
| 11 | Baltimore Ravens | 1542 | 9.6 | 6–13 | |
| 12 | San Francisco 49ers | 1554 | 9.6 | 6–13 | |
| 13 | Pittsburgh Steelers | 1528 | 9.1 | 5–13 | |
| 14 | Green Bay Packers | 1537 | 9.0 | 5–13 | |
| 15 | Los Angeles Chargers | 1540 | 8.9 | 5–13 | |
| 16 | Chicago Bears | 1527 | 8.8 | 5–13 | |
| 17 | Cincinnati Bengals | 1480 | 8.4 | 4–12 | |
| 18 | Tampa Bay Buccaneers | 1479 | 8.2 | 4–12 | |
| 19 | Atlanta Falcons | 1472 | 8.2 | 4–12 | |
| 20 | Kansas City Chiefs | 1485 | 7.9 | 4–12 | |
| 21 | Indianapolis Colts | 1464 | 7.9 | 4–12 | |
| 22 | New Orleans Saints | 1438 | 7.7 | 4–12 | |
| 23 | Dallas Cowboys | 1456 | 7.6 | 4–11 | |
| 24 | Washington Commanders | 1454 | 7.6 | 4–12 | |
| 25 | Cleveland Browns | 1424 | 7.4 | 4–11 | |
| 26 | Miami Dolphins | 1456 | 7.3 | 4–11 | |
| 27 | New York Giants | 1423 | 6.9 | 3–11 | |
| 28 | Carolina Panthers | 1434 | 6.9 | 3–11 | |
| 29 | Arizona Cardinals | 1391 | 6.0 | 3–10 | |
| 30 | Tennessee Titans | 1357 | 5.7 | 2–9 | |
| 31 | New York Jets | 1361 | 5.5 | 2–9 | |
| 32 | Las Vegas Raiders | 1349 | 5.4 | 2–9 |
Benchmark: NumberBench Elo model — K=20, home edge 48, margin-of-victory multiplier, one-third regression to the mean at each season boundary; burn-in on 6,967 NFL regular-season games, 1999–2025. League average rating ≈ 1505. Seattle enters No. 1 as the reigning Super Bowl LX champion.
Every scheduled game, priced
Home-team win probability for all 272 games, priced off entering ratings (see FAQ for why preseason projections work this way). Percentages over 50% mean the home team is favored.
How it’s calculated
Each team’s rating gap to its opponent, plus a 48-point home edge, converts to a win probability via the standard Elo curve (10gap/400 odds form). Season win totals are not computed from those single-game numbers — they come from 5,000 full-season Monte Carlo simulations in which ratings evolve game to game within each simulated season, producing a realistic distribution of outcomes rather than one fake-precise number.
Model output, not a forecast of any specific game’s result — and not betting advice. Ratings will be wrong about specific teams; the ranges say by how much, on average. International/neutral-site games are priced with the listed home team’s standard edge.
Worked example
Week 1: New England (1569) at Seattle (1635). Seattle’s edge is 1635 + 48 − 1569 = 114 rating points, which the Elo curve turns into a 65.8% home win probability — solidly favored, and still losing about one season-opener in three at those odds.
Frequently asked questions
Why is every projection a range and not one number?
The win totals come from 5,000 Monte Carlo season simulations in which team ratings evolve week to week — a team can get hot, cold, or stay steady in each simulated season. The 80% range (10th to 90th percentile) is the honest output: Seattle projecting 11.3 wins with an 8–15 range means preseason knowledge is genuinely that uncertain, which is why single-number season predictions so often look silly by December.
Why does every game use the same team rating all season?
Preseason projections price every scheduled game off each team's entering rating because nobody knows in July which Week 14 starters will be healthy. That is standard practice for season-long projections; game-level numbers get re-priced during the season as information arrives. Treat late-season game probabilities as rough sketches, not forecasts.
How does the Elo model work?
Each team carries a rating (league average ≈ 1505). The gap between two ratings converts to a win probability, the home team gets a 48-point boost, and after real games the winner takes rating points from the loser — more for upsets and blowouts (K = 20 with a margin-of-victory multiplier). Ratings regress one-third of the way to the mean each offseason. The model was burned in on 6,967 real regular-season games from 1999–2025.
Why are the defending champions on top?
Seattle enters 2026 with the No. 1 rating (1635) because ratings are earned on the field, and the Seahawks won Super Bowl LX. The model then regresses everyone toward the mean — which is why their projection is 11.3 wins, not a repeat of last season's total.
Can a team projected under .500 make the playoffs?
Easily. A 7.9-win projection with a 4–12 eighty-percent range means roughly a fifth of that team's simulated seasons land at 10+ wins. Projections describe the center of a distribution, not a ceiling.
What about the Melbourne game and other neutral sites?
The 49ers–Rams opener in Melbourne, Australia is priced with the listed home team's standard 48-point edge, like every other game. Neutral-site and international games are a known simplification in preseason models — the true home edge there is likely smaller.