The win-total board.

Every team's season, simulated 10,000 times off preseason Elo priors — wins mean, the 10th-to-90th percentile band, division and playoff odds. Market total, de-vigged implied, and the sim-vs-market edge join the moment a market win-total feed exists; edges are labeled leans, never picks.

TEAMS32SIMS10,000VALIDATION2015-2025 · MAE 2.338MARKETMODEL-ONLYUPDATED9m ago

The survivor primer The gate report

SCORED ON SEASONS IT NEVER SAW

2.338 MAE vs 2.813 / 2.602

Walk-forward over 2015-2025: the sim's win projections against two baselines — last season's wins (2.813) and a flat 8.5 for everyone (2.602). Lower is better. It beat both.

ALL 32 TEAMS

Projected wins, with the band.

MODEL-ONLYNo market win-total feed exists from either market source probed — one odds aggregator and one box-score provider (checked every cycle since 2026-08-11). The board runs model-only; market total, implied, edge, and lean columns publish the day a feed posts. No reconstructed numbers.

SORT

TEAM	WINS	P10–P90 BAND	DIVISION	PLAYOFFS	MARKET TOTAL	MODEL P(OVER)	EDGE · LEAN	ELO PRIOR
Seattle Seahawks SEA	12.3	10–15	71.8%	96.9%	—	—	—	1675.8
Houston Texans HOU	11.1	9–14	66.8%	89.8%	—	—	—	1603.5
Denver Broncos DEN	11.0	9–13	75.7%	89.7%	—	—	—	1613.4
Los Angeles Rams LA	10.7	8–13	20.5%	83.8%	—	—	—	1610.7
Buffalo Bills BUF	10.6	8–13	58.9%	85.5%	—	—	—	1610.3
Philadelphia Eagles PHI	10.4	8–13	78.7%	85.2%	—	—	—	1571.2
New England Patriots NE	10.3	8–13	37.9%	79.5%	—	—	—	1594.8
Detroit Lions DET	10.3	8–13	47.4%	75.8%	—	—	—	1559
Jacksonville Jaguars JAX	10.1	8–13	28.1%	74.6%	—	—	—	1566.2
Baltimore Ravens BAL	10.0	8–13	56.9%	73.9%	—	—	—	1547.5
Minnesota Vikings MIN	9.8	7–12	30.2%	64.0%	—	—	—	1552.7
San Francisco 49ers SF	9.5	7–12	7.6%	61.3%	—	—	—	1553.5
Pittsburgh Steelers PIT	8.9	6–11	23.5%	47.0%	—	—	—	1512.3
Green Bay Packers GB	8.9	6–11	13.2%	42.8%	—	—	—	1530.7
Los Angeles Chargers LAC	8.6	6–11	14.0%	42.9%	—	—	—	1525.5
Chicago Bears CHI	8.6	6–11	9.2%	34.4%	—	—	—	1526.9
Cincinnati Bengals CIN	8.5	6–11	14.9%	34.5%	—	—	—	1476.3
Kansas City Chiefs KC	8.3	6–11	10.1%	32.2%	—	—	—	1500.5
Tampa Bay Buccaneers TB	8.3	6–11	43.7%	47.2%	—	—	—	1479.7
Atlanta Falcons ATL	8.0	5–11	30.3%	35.3%	—	—	—	1465.2
Indianapolis Colts IND	7.8	5–10	4.8%	22.3%	—	—	—	1464.1
New Orleans Saints NO	7.6	5–10	19.4%	23.9%	—	—	—	1430.4
Washington Commanders WAS	7.3	5–10	9.5%	17.0%	—	—	—	1453.7
Dallas Cowboys DAL	7.2	5–10	7.9%	14.3%	—	—	—	1452.5
Cleveland Browns CLE	7.2	5–10	4.7%	12.6%	—	—	—	1411.9
Miami Dolphins MIA	7.0	5–10	2.9%	11.8%	—	—	—	1452.7
New York Giants NYG	6.5	4–9	3.9%	7.2%	—	—	—	1413.7
Carolina Panthers CAR	6.4	4–9	6.7%	9.0%	—	—	—	1420.5
Arizona Cardinals ARI	5.5	3–8	0.1%	1.8%	—	—	—	1385.6
Tennessee Titans TEN	5.2	3–8	0.3%	1.5%	—	—	—	1343.7
New York Jets NYJ	5.1	3–7	0.3%	1.3%	—	—	—	1352.6
Las Vegas Raiders LV	4.8	3–7	0.3%	0.9%	—	—	—	1342.8

METHOD · CONSTANTS · CAVEATS

How the board is built.

METHOD

Preseason Elo built from every result since 1999 (regressed one-third toward league average each offseason), home field estimated from the last ten completed seasons. 10,000 seeded simulations of the regular season with ratings frozen at preseason priors. Validated out-of-sample: every season in the window replayed from priors only and compared to actual win totals.

SIMS 10,000 · SEED 20260907 · ELO START 1500 · K 20
MOV ×ln(1+margin) · REGRESSION one-third toward 1500
HFA +33 (home win% 54.7% over 2016-2025 (non-neutral)) · TIEBREAK simulated standings ties broken by Elo — real NFL tiebreakers not modeled

Win distributions from the season sim (10,000 seeded simulations off preseason Elo priors — see the sim feed for constants and validation). Where a market total exists: model P(over) is the share of simulations finishing above the total (pushes excluded), market implied is the two-way book de-vigged by normalization, edge_pp = model − market in percentage points.

VALIDATION RULER · 2015-2025

Season sim2.338 MAE

Prior-year wins2.813 MAE

Flat 8.5 wins2.602 MAE

Projected win totals beat both naive baselines (prior-year wins and flat .500) out-of-sample.

validated

LEANS, NOT PICKS

OVER/UNDER flags are LEANS, posted at |edge| ≥ 3pp against the de-vigged market — never picks. The NFL sides model failed its gate, and no win-total lean becomes bet language until a graded, market-relative record exists. Sizing, when anything ever clears, is in units (1u = 1% of budget), never dollars.

Priors are a starting point, not a pick.

The sim says where teams start, not where they finish — the in-season re-rating reads real games as they're played. The weekly side of the vertical is gated until a model beats the close, and win-total leans stay leans until a graded, market-relative record exists.

The survivor primer The gate report