Kalshi NFL Week 3 — Contract Prices vs. Our Model, All 16 Games
Updated · contract prices
Every Week 3 game with three numbers on one line: our model's win probability, the live Kalshi contract price for that same side, and the gap between them. A game contract pays $1 if the team wins, so the price in cents is the market's implied probability — the two are directly subtractable. Biggest disagreements first.
Quick Answer
As of September 28, 2026, the widest Week 3 disagreement between our model and Kalshi is Philadelphia Eagles at 21.7 percentage points — the model has that side at 62% where the contract is asking 40.5¢.
1 of 16 games carry a live contract price; Kalshi opens game markets as kickoff approaches.
Model and Kalshi columns are both the home side, so a negative edge is value on the away team — the row names it.
The model is DAEPA: 59.6% straight up across 3,266 walk-forward games, Brier 0.236.
Free Week 3 edge sheet — all 16 games, our number vs the Kalshi contract price.
No spam. Unsubscribe anytime.
Model and Kalshi columns are both the home side, so the edge is home-perspective; a negative edge is value on the away team, named in the row. 1 of 16 games carry a live contract price — Kalshi opens game markets as kickoff approaches, and an unopened market shows an em-dash rather than an estimate.
Ranked by disagreement: the live NFL edge board
The table above is every game in model-line order. This board ranks the current week by how far the model and the Kalshi price disagree, largest gap first, and rotates every 5 minutes.
Where else to look
Kalshi NFL Markets is the season-long anchor — futures, division and championship contracts alongside the weekly board. Week 3 predictions carries the same games with the full model write-up per matchup. Power ratings vs. the market does the same comparison at the season level, in rating points.
Kalshi NFL Week 3 — questions
How do I read a Kalshi NFL Week 3 contract price?
A Kalshi game contract pays $1 if that team wins and $0 if it does not, so the price in cents is the market’s implied probability directly — a contract at 62¢ is the market saying 62%. That is what lets us subtract our model’s probability from it and state the difference in percentage points, with no odds conversion in between.
What does the edge column mean?
Edge is our model’s win probability for a side minus the market’s price for that same side, in percentage points. Positive means the model thinks the market is underpricing that team. It is a disagreement, not a guarantee — every contract can settle at $0.
Where does the model number come from?
The Gridiron Edge engine, built on Defense-Adjusted EPA — per-play efficiency adjusted for opponent quality. It grades at 59.6% straight up across 3,266 walk-forward games (2013–2024), Brier 0.236. Those figures grade DAEPA itself, not any composite built on top of it. Every number here is published before kickoff and graded afterwards.
Is every Week 3 game priced yet?
Not always. Kalshi opens game markets as kickoff approaches, so early in a week some games have a model number and no contract price. Those rows show an em-dash in the price and edge columns rather than an estimate — an unopened market has no price, and inventing one would be the whole problem.
Go Pro · 26% off your first year
Every NFL game priced against the live Kalshi line, every week — $150/yr
Pro is the layer around the free game edges: fair value against the live Kalshi price on every market, edge alerts in Discord the moment a gap opens, and the full win-total, championship and projection boards.
$111 for the first year, then $150/yr. Code applies automatically. Ends Dec 31. Monthly is $14.99/mo.
Every signal graded in public, losers left on the board. See the record →
More NFL Markets
- NFL Week 6 predictions — Model vs. Kalshi on every game on the board right now.
- NFL Week 7 predictions — early lines — Next week’s slate, priced the moment the lines open.
- NFL Week 5 results — Final scores and every graded call from the week just played.
- Markets 101: be the market maker, not the market taker → — Why the first-TD board prices out to ~118%, and how a resting bid a cent under the ask beats paying it
- Heisman odds 2026 — Kalshi vs Polymarket → — Every contender priced on both exchanges, cheaper venue named, plus Coach Gene’s longshots
- Grok fantasy football — connect the draft desk → — One URL, no key: the grok.com/connectors walkthrough and six prompts
- NFL MCP server — the in-season desk, inside your AI agent → — Kalshi props venue by venue, spread ladders, correlation-aware combos, weekly game edges — free tools, no key
- MCP setup — connect Claude, ChatGPT or Grok in a minute → — One URL, no API key: per-host install steps and troubleshooting
- NFL 2026 Market Guide — the 61-page book → — All 32 teams priced against the market, plus a 153-player draft board
- NFL 2026 Quarter-Season Report — 64 games, 32 teams, priced → — Weeks 1–4 in one book: opening vs closing lines, the Kalshi board Sept 8 vs Oct 8, all 32 team cards
- NFL Player Props vs Kalshi — the edge board → — Our projection vs the live Kalshi prop line, biggest gap first
- Madden 27 ratings vs our model vs the market → — All 32 team ratings, our power ratings, and live win-total prices on one board
- NFL Predictions — every game vs the market → — Free fair value vs Kalshi on all 18 weeks
- 2026 NFL Win Totals — model vs Kalshi → — All 32 teams: projected wins vs the KXNFLWINS ladder
- NFL Standings 2026 — projected records + playoff odds → — All 8 divisions: projected record, PWR, division-title and playoff odds
- NFL Schedule Predictor 2026 — every game picked → — Straight-up winner and win probability for all 272 games, plus the 10,000-season sim
- NFL Game Edges — where the model disagrees → — Moneyline, spread & total edges vs Kalshi prices
- NFL MVP Edges → — Live MVP market mispricings vs Kalshi KXNFLMVP
- Win Total Futures → — Monte Carlo season wins vs the Kalshi line
- Championship Edges → — Playoff / conference / Super Bowl probabilities
- Kalshi NFL Markets → — How Kalshi prices every NFL market + live edges
- Power Rankings — the model methodology anchor → — All 32 teams on the points-per-game scale
- Fantasy Draft Assistant — the same model, in your AI agent → — Free MCP draft desk: board, best available, sleepers vs ADP, who-do-I-draft
- NFL Model Calibration Ledger — how accurate is it, actually → — The graded record: 59.6% straight up, 0.236 Brier, and hit rate by confidence bucket across 3,266 games
- NFL Power Ratings tool — build your own PWR board → — Adjust the model weights and re-rank all 32 teams on the points-per-game scale
- NFL Win Probability calculator → — Turn any rating gap, spread or moneyline into a calibrated win probability
- NFL Power Ratings vs. the Market — where we disagree → — Our PWR and the rating implied by live Kalshi win-total prices, all 32 teams on one points scale
Win probabilities are the market-anchored fair value once a game is priced; the model’s own line is the lean. Prediction markets carry risk; contracts can settle at $0. Informational only — trade responsibly.