2026 NFL Win Totals: Model vs Kalshi Markets
Projected wins for every 2026 NFL team — 272 regular-season games played out one matchup at a time through our power-rating + home-field win-probability model — compared head-to-head with the live Kalshi KXNFLWINS-27 17-rung strike ladder. Our headline projection blends the model halfway toward the market in the preseason (a humility anchor that fades to zero by Week 6 as real results arrive). The Edge column shows the model’s full disagreement with the market — raw projected wins minus the market — and we flag OVER / UNDER when that gap is at least 2.50 wins (a bar that eases toward 1.25 as the season validates the model).
Want the reasoning behind the numbers rather than just the board? Coach Gene Clemons and Chris James walk five of the widest gaps on this table — the Jets, Dolphins, Cardinals, Browns and Raiders — with the Raiders the only fade of the five. For a completely independent read on the same 32 teams, the Madden 27 ratings board simulates a win total for every club from EA's launch ratings alone.
How We Project Season Wins
We use the same 2026 power ratings that drive our Week 1 slate and power rankings. For each of a team's 17 regular-season games we compute a single-game win probability — power-rating differential plus a 2.5-point home-field bump, converted to a probability through a Normal margin distribution with σ=13.5 — and then sum those 17 probabilities to get expected wins. Variance follows from Poisson-binomial summation (Σ p × (1−p)), not a Normal approximation, because the per-game probabilities are independent but not identically distributed.
On the market side, Kalshi's KXNFLWINS-27 series is a 17-rung survival ladder — separate YES contracts for “≥1 wins,” “≥2 wins,” through “17 wins.” For any non-negative integer random variable, expected value equals the sum of survival probabilities, so the market's implied expected wins is simply the sum of the 17 mid prices. No de-vigging needed — Kalshi is binary YES/NO.
Because a preseason model is a prior with no in-season signal yet, our headline projection is a blended one — w·model + (1−w)·market, with the model weight w starting at 0.5 in the preseason and ramping to 1.0 by Week 6 as real results validate the model. The Edge column, though, shows the model’s full disagreement with the market — raw projected wins minus the market — so nothing is hidden. We recommend OVER when that raw gap is at least 2.50 wins above the market, UNDER when it’s 2.50 below — a deliberately high bar in the preseason that eases toward 1.25 wins as the blend weight climbs to 1.0 and the model earns trust. The recommended strike is whichever rung trades closest to 50¢ — the most informationally efficient point on the ladder.
All 32 Teams — Projected vs Market
Tap any column to sort. Default: largest absolute edge first.
| Team | PR | Proj. Wins | Market E[W] | Edge = model−mkt (|Δ|≥2.50) ↓ | Verdict | Action |
|---|---|---|---|---|---|---|
 DolphinsBelow Avg | -3.19 | 5.51 model 6.63±1.96 | 4.38 | +2.25 | Pass | View → |
 CardinalsBelow Avg | -3.76 | 5.36 model 6.35±1.94 | 4.36 | +1.99 | Pass | View → |
 BrownsBelow Avg | -2.11 | 6.80 model 7.79±2.02 | 5.82 | +1.97 | Pass | View → |
 VikingsBelow Avg | -2.46 | 8.06 model 7.11±1.99 | 9.01 | -1.91 | Pass | View → |
 BengalsAverage | -0.56 | 9.36 model 8.52±2.03 | 10.20 | -1.68 | Pass | View → |
 ChiefsAverage | -0.54 | 9.01 model 8.22±1.98 | 9.79 | -1.57 | Pass | View → |
 FalconsAverage | -0.32 | 7.72 model 8.50±2.04 | 6.93 | +1.56 | Pass | View → |
 ChargersAverage | -0.65 | 8.76 model 8.05±1.98 | 9.46 | -1.41 | Pass | View → |
 PatriotsContender | +5.30 | 10.37 model 10.94±1.91 | 9.80 | +1.14 | Pass | View → |
 RavensPlayoff | +1.40 | 9.82 model 9.29±2.01 | 10.35 | -1.05 | Pass | View → |
 RaidersRebuild | -6.16 | 5.89 model 5.47±1.89 | 6.31 | -0.84 | Pass | View → |
 TexansPlayoff | +1.95 | 9.71 model 9.31±2.02 | 10.11 | -0.80 | Pass | View → |
 RamsContender | +5.18 | 11.12 model 10.73±1.95 | 11.52 | -0.79 | Pass | View → |
 49ersPlayoff | +1.11 | 9.38 model 8.98±2.00 | 9.77 | -0.79 | Pass | View → |
 EaglesPlayoff | +2.06 | 9.79 model 9.41±2.02 | 10.18 | -0.77 | Pass | View → |
 BroncosPlayoff | +1.24 | 9.49 model 9.11±2.01 | 9.88 | -0.77 | Pass | View → |
 LionsPlayoff | +2.60 | 10.25 model 9.92±1.98 | 10.58 | -0.67 | Pass | View → |
 PanthersAverage | -1.11 | 7.53 model 7.86±2.03 | 7.20 | +0.66 | Pass | View → |
 CowboysPlayoff | +1.90 | 9.34 model 9.04±1.99 | 9.65 | -0.60 | Pass | View → |
 BearsPlayoff | +1.47 | 9.14 model 8.85±2.00 | 9.43 | -0.58 | Pass | View → |
 JetsBelow Avg | -4.12 | 6.15 model 6.43±1.94 | 5.87 | +0.56 | Pass | View → |
 ColtsAverage | +0.00 | 8.36 model 8.61±2.03 | 8.12 | +0.50 | Pass | View → |
 SteelersAverage | -0.47 | 8.43 model 8.20±2.02 | 8.65 | -0.46 | Pass | View → |
 PackersPlayoff | +3.14 | 9.58 model 9.80±1.99 | 9.36 | +0.44 | Pass | View → |
 SeahawksContender | +5.63 | 10.86 model 11.06±1.93 | 10.65 | +0.41 | Pass | View → |
 JaguarsPlayoff | +2.69 | 9.29 model 9.44±2.01 | 9.14 | +0.30 | Pass | View → |
 CommandersAverage | -1.09 | 7.49 model 7.63±2.03 | 7.35 | +0.28 | Pass | View → |
 TitansBelow Avg | -3.58 | 6.53 model 6.67±1.97 | 6.40 | +0.26 | Pass | View → |
 SaintsBelow Avg | -2.28 | 7.83 model 7.73±2.00 | 7.92 | -0.20 | Pass | View → |
 GiantsAverage | -1.08 | 7.61 model 7.69±1.98 | 7.52 | +0.17 | Pass | View → |
 BillsContender | +4.28 | 10.42 model 10.37±1.95 | 10.47 | -0.10 | Pass | View → |
 BuccaneersAverage | -0.50 | 8.27 model 8.27±2.02 | 8.27 | -0.00 | Pass | View → |
Methodology Notes & Outliers
Kansas City sits at PR -0.54 (rank 19) on the engine-pure preseason prior — no manual Mahomes override. We retired that patch: availability now reads live from the injury feed at edge-compute time instead of freezing a health guess into the rating, so KC's number moves with real news rather than a static flag.
Baltimore at PR +1.40 will surprise. The 2026 preseason prior shrinks heavily toward league mean for teams coming off volatile late-season runs. The early-season prior is the starting point — by Week 4 in-season DAEPA blends in and ratings re-stratify quickly.
New England and Seattle at PR +5.30 / +5.63 top the board — both teams the market currently has well outside the projected-wins leaderboard. Treat the magnitude with appropriate skepticism: power-rating priors compress to the mean, but the relative ranking is more reliable than the absolute number. For the full Seattle case — why the defending champs sit at ~10.9 projected wins against a 10.5 market line — see Coach Gene's Seahawks win-total deep-dive.
Poisson-binomial vs Normal: A 17-game schedule with heterogeneous win probabilities is exactly the Poisson-binomial case. We sum p × (1−p) for variance directly rather than fitting a Normal — at season totals around 8–10 wins with mixed-strength opponents the difference is small but the math is honest.
Want the full methodology? See our Gridiron Edge model documentation — empirical per-component shrinkage, 27-year YoY stability fit, and a 2.32-game season win-total MAE out of sample.
More NFL Coverage
- 2026 Week 1 Slate — model vs Kalshi for all 16 opening games
- 2026 Power Rankings — all 32 teams on the points-per-game scale
- Methodology — DAEPA, priors, win-probability conversion
- NFL Hub — power rankings & weekly edge preview
- Kalshi Markets Hub — live prediction-market movers
- Sports Markets — NFL, NBA, soccer, World Cup 2026
Drafting this month? Same model, fantasy side.
The 2026 projection engine behind these numbers also runs a free fantasy draft desk inside Claude, ChatGPT or Cursor — the ranked board, the best available at your pick, sleepers and busts vs consensus ADP, and a who-do-I-draft call mid-draft. 8 tools, no account.
Open the fantasy draft assistant →NFL win totals as prediction markets — FAQ
How are NFL win totals priced as prediction markets?
Kalshi’s KXNFLWINS series lists a ladder of yes/no contracts for each team — "wins ≥ 8.5", "wins ≥ 9.5", and so on. Each rung’s price is the market-implied probability the team clears that win threshold. Reading the ladder gives an expected-wins number, which is what we compare against our model.
What does the model do that the win-total market does not?
Our DAEPA power-rating model projects every team’s 17-game schedule, turning each matchup into a win probability and summing them to an expected-wins total. We then set that against the Kalshi ladder and flag any team where the raw model−market gap clears our edge bar — about 2.5 wins in the preseason, easing to 1.25 by Week 6 as results validate the model — the point where the gap is worth a position rather than noise.
Can you trade NFL win totals on Kalshi?
Yes. Kalshi is a CFTC-regulated exchange available in all 50 states. You take a yes/no position on a team clearing a win threshold and it settles at $1 or $0 at season’s end — a contract, not a wager against a sportsbook line.
When do 2026 NFL win totals settle?
Each team’s win-total contracts settle once their 2026 regular season is complete — after Week 18. The ladder prices drift all season as results come in; an over rung that looked like a coin flip in September can settle early if a team clinches the number with games to spare.
What is an NFL win total?
A win total is the number of regular-season games a team is projected to win across its 17-game schedule. On a sportsbook it’s a single over/under line with vig baked into both sides. On a prediction market like Kalshi it’s a ladder of yes/no contracts — "wins ≥ 8", "wins ≥ 9", and so on — where each rung’s price is the market-implied probability of clearing that threshold, no vig.
How do I read the Kalshi KXNFLWINS ladder to get an expected win total?
For any non-negative whole-number outcome, expected value equals the sum of the survival probabilities — so a team’s market-implied expected wins is simply the sum of its rung prices across the KXNFLWINS-27 ladder. Add up the "≥1", "≥2", … "≥17" mid prices and you have the market’s expected-wins number, which is exactly what our model’s projected wins is compared against on this page.