GridironProjector NFL
EPA-driven NFL score projections
Opponent-adjusted points and EPA ratings from nflverse play-by-play, shrunk toward preseason priors, with a quarterback value system, rest, travel and weather. Home field is about 2.6 points before travel, fitted on past seasons. Win chances are calibrated on history. How it works and its track record.
🎯 Hit rate
At 80%+ a signal needs 20+ graded games unless you pick a sample (tap the auto chip to remove it).
Matchup presets
- 🐢 Slow game · Leans Under: fewer points than the total
- Both offenses rank in the bottom 10 for plays per game (pace). Fewer snaps means fewer chances to score for both teams, so totals tend to come in low.
- 🛡️ Defensive battle · Leans Under: fewer points than the total
- Both defenses rank top 8 in EPA per play allowed. Two defenses that stop plays efficiently cut each other's scoring.
- 💥 Shootout · Leans Over: more points than the total
- Both offenses top 8 in explosive play rate, or both defenses bottom 10 in EPA allowed. Big-play offenses or leaky defenses at both ends put points up quickly.
- 🧱 Red-zone stall · Leans Under: fewer points than the total
- Both offenses bottom 10 in red-zone touchdown rate: drives end in field goals. Drives that stall inside the 20 end in field goals (3) instead of touchdowns (7), shaving the total.
- 🚧 Red-zone walls · Leans Under: fewer points than the total
- Both defenses top 8 in red-zone TD rate allowed. Both defenses force field goals near the goal line, so long drives score fewer points.
- 🏎️ Track meet · Leans Over: more points than the total
- Both offenses top 10 in pace and top 14 in EPA per play: lots of snaps, used well. Many snaps used efficiently by both offenses means a high-scoring game.
- 🎯 Air raid vs soft secondary · Leans Over: more points than the total
- An offense top 6 in pass EPA per dropback faces a defense bottom 10 in pass EPA allowed. An elite passing game against a weak secondary produces chunk plays and quick scores.
- 🏁 Both finish drives · Leans Over: more points than the total
- Both offenses top 10 in points per drive. Both offenses turn possessions into points, so every drive is a scoring chance.
- 🏟️ Dome shootout · Leans Over: more points than the total
- Played indoors (dome or closed roof) and both offenses top 16 in EPA per play. No wind, rain or cold, and two above-average offenses: indoor games with good offenses run high.
- 🔨 Mismatch · Leans favorite covers the spread
- Favorite's offense top 6 in EPA per play vs an underdog defense bottom 10 in EPA allowed. A strong offense against a poor defense tends to keep scoring after the game is decided.
- 🧹 Complete mismatch · Leans favorite covers the spread
- Favorite top 4 in EPA per play and top 8 in EPA allowed; underdog offense bottom 10 in EPA per play. The favorite is better on both sides and the underdog can't move the ball, so margins run large.
- 🐶 Dog with a defense · Leans underdog covers the spread
- Underdog defense top 8 in EPA allowed vs a favorite offense bottom 12 in EPA per play. A good underdog defense against a weak favorite offense keeps the game close.
- 🏠 Home dog, better D · Leans underdog covers the spread
- Home underdog whose defense ranks 5+ spots better in EPA allowed than the favorite's. Home field plus the much better defense tends to keep a home underdog inside the number.
- 🐺 Live dog · Leans underdog wins outright
- Underdog ranks 3+ spots better than the favorite in both EPA per play and EPA allowed. The underdog grades better on both offense and defense, so it can win outright.
- 🤝 Divisional dog · Leans underdog covers the spread
- Division game, and the underdog's defense ranks top 16 in EPA allowed. Division rivals play twice a year and know each other's schemes; a dog with a real defense tends to keep it close.
- 🛌 Rest edge · Leans the team with the edge (named on each game) to cover
- One team has 3+ more days of rest (off a bye, or a Monday game into a short week for the other). The rested side covers. Extra days heal bumps and add prep time; a short week cuts both.
- 🎁 Turnover edge · Leans the team with the edge (named on each game) to cover
- A defense top 8 in takeaways per game faces an offense bottom 10 in giveaways. The takeaway side covers. Takeaways flip field position and possessions toward the side that forces them.
- 🌪️ Pressure edge · Leans the team with the edge (named on each game) to cover
- A pass rush top 6 in pressure rate faces a line bottom 10 in pressure rate allowed (NFL stand-in for college havoc). Pressure blows up dropbacks, forces bad throws and creates turnovers for the rushing side.
- 💨 Sack mismatch · Leans the team with the edge (named on each game) to cover
- A pass rush top 6 in sack rate faces a line bottom 10 in sack rate allowed. Sacks kill drives and force long third downs for the team that can't protect.
- 🏆 Finishing edge · Leans the team with the edge (named on each game) to cover
- An offense top 6 in red-zone TD rate faces a defense bottom 10 in red-zone TD rate allowed. Turning red-zone trips into touchdowns, against a defense that can't stop them, widens the margin.
- 🪨 Run game into a wall · Leans the team with the edge (named on each game) to cover
- A run defense top 6 in rush EPA allowed faces an offense bottom 10 in rush EPA. The defense's side covers. A run game that can't move the ball puts the offense behind the chains all day.
- 📊 Study
- Multi-season studies: ⭐ sharp patterns (2010–25, beat the closing price out of sample), stat trends and tail leans (2016–25 with a 2025 holdout), total combos (2016–24 + 2025 holdout) and long-run 💎 diamonds. Strength: Strong = sharp pattern, super trend or long-run diamond; Moderate = watching trend or total combo; Lean = sharp lean or tail lean. 🐶 Underdog spots = Study badges backing the underdog to win outright.
- 🧩 2026 combo
- This season only, searched again every week. Held up = a this-season 💎 diamond (cleared the luck check); Weak = a 🔥 hot-this-season combo or 2026 streak chip (could be luck). Small samples.
- 🔥 Undefeated · 💎 Near-perfect · ⭐ Elite
- Found automatically on 2016–2021 (1548 NFL games with a line): every one- and two-part combination of both teams' frozen pregame ranks and matchup edges, the line and total, our number vs the line, rest, division, venue, roof, kickoff weather, season stage and each team's last result and cover. Kept only when it hit 75%+ on enough games that shuffled results rarely do it by luck, beat what the line itself implies, and then held up on 2022–2025. 🔥 no losses, 💎 90%+; ⭐ 72%+ and 10+ points better than usual on games it never saw (20+ of them, 40+ in all). 2026 is the untouched test record on each pill. On shuffled results the whole search finds 0 per run by luck; on the real games it kept 0. Player conditions (🔥 / 💎 / ⭐ on player pills) use the same guards on 2021–2025 player games, tested on 2026.
- 🎯 Hit rate
- One bar for every pill with a record (top of the board). Applies to: All signals, or pick types (Team: Spread / Moneyline / Total by the pill's outcome, Players, Patterns, Study, Combo, Elite). Several types: any one is enough, or All must match, each with its own bar. Edge over typical = record minus what that outcome usually does on those games (the line's implied win chance for a moneyline call). Matching pills move to the top of the card in a ring; the rest are dimmed.
- Side
- Keep only Study, combo and undefeated badges pointing that way (Over, Under, Favorite = covers or wins, Underdog = covers or wins). Cards then show only the badges that pass. The minimum sample now sits in 🎯 Hit rate (n≥20 automatically at 80%+).
- 🤝 Agrees with our model · 🔗 2+ agree
- Agrees: our projection is on the same side against DraftKings (Over and our total above the book's; Favorite covers and our margin above the spread). 2+ agree: at least two badges on the card point the same way (Study, combo, undefeated, pattern alerts, matchup alerts).
- 🎯 Each combo · Best record sort
- Pick any named combo to see only its games (tap a 🎯 name on a card to do the same). Best record ranks by the lower end of each badge's 95% range (Wilson), so 219-142 beats 4-0. Records only, not advice. 21+.
Ranks are among the 32 NFL teams (1 = best; for defensive stats 1 = fewest allowed), from games played before each week's kickoffs. Records grade this season's finished games against the closing line. Tap a pill on any game to see both teams' actual ranks.
Badges, patterns and results
This season only — small sample. Diamonds come from this season's games alone, searched again every week: 8+ games, 12+ points above the no-vig closing price, held up on the latest two weeks, and better than a luck check on simulated seasons. A diamond that stops holding drops off the next week. 🔥 Hot this season = a board tag (Shootout, Low scoring, Coin flip, a stat-trend chip, a matchup alert...) plus one or two more pregame factors that cleared the same bar but not the luck check: this-season runs that could be luck (at most two per card, the tag first). This season's diamonds and their records · Every hot combo
Pattern tags
Situations that beat games on similar spreads in our 2020–25 backtest. How they're graded
Stat trends
Team-stat matchups that beat games on similar lines, 2016–25, with a 2025 holdout. 🎯 Tail leans are experimental small-sample rules (may be noise). A result gets a chip only when it has at least one rule. Records and what failed
Player outlook
Games with a player badge. Each badge compares a player's projection with his own season average.
Player success rate: set it in 🎯 Hit rate at the top of the board (Applies to: Players).
Low passing yards, High receiving yards, High receptions, Receptions leader, TD threat didn't clear 55% on 2025–26 games, so they're not shown. How it's tested
Matchup alerts
Tap any to keep only the games carrying those badges.
Slider filters
Our total minus the DraftKings total. +3 means we project 3 points more than the book.
Drag either end to keep only the games whose numbers land inside every range you moved. A handle left at its end means no limit on that side, and a slider you haven't touched never hides a game. While a slider is set, games without that number (no projection yet, no DraftKings line) are left out. Our numbers are the pregame projection, also on live and final cards.
Chip counts and the list follow these settings: a chip counts a game only when the badge it matches is shown on the card.
Loading the schedule…
NFL Week 8 at a glance
Week 8 of the 2026 NFL season has 14 games on our board. The closest projected matchup is Vikings at Lions (0-point margin), the highest projected total is Cardinals at Cowboys (54.7 points) and the lowest is Browns at Steelers (40.4). 3 games are rated a toss-up (neither side above 58%).
- Vikings at Lions — We project Lions 23-23, a 0-point margin with 46.8 total points; Lions win 50% of the time.
- Raiders at Jets — We project Jets 23-23, a 0.6-point margin with 46.2 total points; Jets win 52% of the time.
- Chiefs at Broncos — We project Broncos 22-20, a 1.6-point margin with 42.2 total points; Broncos win 55% of the time.
- Eagles at Commanders — We project Commanders 24-20, a 3.2-point margin with 43.8 total points; Commanders win 60% of the time.
- Bears at Seahawks — We project Seahawks 25-21, a 4-point margin with 45.5 total points; Seahawks win 62% of the time.
- Falcons at Buccaneers — We project Buccaneers 24-20, a 4.2-point margin with 44.2 total points; Buccaneers win 63% of the time.
All Week 8 projections · NFL power rankings · Playoff predictions
Projections are statistical estimates for entertainment and analysis only. They are not betting advice and nothing here guarantees any result.
Free weekly projections email
One email a week in season: the closest projected games, the biggest mismatches and how the model graded out last week. Email address only — no spam, unsubscribe any time.