BALOut
Lamar Jackson ruled out
Misses the game against the Browns (ankle). That costs the Ravens 6.5 points in the model.
Thu Sat out → Fri Sat outBAL @ CLE →
The model picked 0 of 1 winners, the same as the market. One week is far too few games to show an edge, so read this as a progress report.
The model's most confident pick, its best chance of an upset, the tightest game and where it disagrees most with the market.
The model's most confident pick, over the Cardinals. The market has it at 85%.
ARI @ LA · Mon 19 Oct, 6:05am Upset watch56%The market makes the Browns the underdog at 40%. The model gives them 56%, so it picks the upset.
BAL @ CLE · Mon 19 Oct, 3:00am Closest call52%As near to a coin flip as the week gets. The model leans to the Seahawks by 0.7.
SEA @ DEN · Final 14–17 Biggest split with the market5.1Model: Browns −2.1. Market: Ravens −3.
BAL @ CLE · Mon 19 Oct, 3:00amWin chance for each side, the model's score and the market line. Finished games show the final score. Tap a game for its full breakdown.
The Seahawks were the stronger team last season. The Broncos are missing more players this week. The bookies are more confident (58% there).
The Jaguars have played better this season. The Jaguars were the stronger team last season. The model is more confident than the bookies (57% there).
The Falcons are at home. The Falcons have the better quarterback. The bookies have this the other way: they favour the Bears (52%).
The Ravens are missing more players this week. The Browns are at home. The bookies have this the other way: they favour the Ravens (60%).
The Colts were the stronger team last season. The Colts have played better this season. The bookies see it about the same (74%).
The Patriots were the stronger team last season. The Patriots have played better this season. The model is more confident than the bookies (74% there).
The Giants are at home. The Giants have played better this season. The model is more confident than the bookies (50% there).
The Eagles are missing more players this week. The Panthers have played better this season. The model is more confident than the bookies (60% there).
The Buccaneers are missing more players this week. The Steelers were the stronger team last season. The bookies see it about the same (57%).
The Rams were the stronger team last season. The Rams have played better this season. The bookies see it about the same (85%).
The Chiefs have played better this season. The Chiefs are at home. The bookies are more confident (78% there).
The Bills were the stronger team last season. The Bills have played better this season. The bookies see it about the same (67%).
The Packers are at home. The Packers were the stronger team last season. The model is more confident than the bookies (51% there).
The 49ers have played better this season. The 49ers were the stronger team last season. The bookies see it about the same (78%).
Times are Brisbane (AEST). The bar splits at each side's chance of winning, in team colours, away side on the left.
The model's yards for each player next to his average this season, biggest differences first. That gap is where the model and a bookie's line are most likely to disagree. Type the bookie's line to see the model's chance of the over and the under.
Regular starters with three or more games this season (at least 20 rushing or receiving yards a game, or 120 passing), in games still to play. Sorted by the gap measured against how much that player's yards usually swing. The model's yardage numbers were tested on 2022 to 2025 seasons against what players actually did, but not yet against bookmakers' lines, so treat a big gap as a reason to look, not a sure thing.
Winning margin for the home side, in points. The further apart the two dots, the more the model disagreed with the market. Widest gaps first.
Combinations priced by the model. The fair price is the odds that would break even if the model is right; a bookmaker will offer less.
Longer odds than a single bet, and every extra leg is another chance to lose.
These are the games where the model and market disagree. The market is usually the better judge, so the Prediction Log is the test of whether this is an edge or noise.
Touchdown scoring is high-variance. Treat as entertainment, not a strategy.
Every extra leg multiplies the chance of losing, and a bookmaker's margin is built into each leg. These are probabilities from a model, not tips. If gambling is a problem for you or someone you know, Gambling Help is on 1800 858 858.
Points of margin each absence cost his team this week.
Out and Doubtful count as 0% available. Questionable counts as 50%, 65% or 75% depending on Friday practice, from how often such players took the field in 2023–25. Players with no game status are treated as playing.
Chance each player scores a rushing or receiving touchdown, with the price that would be fair.
Back-tested on the 2025 season: players rated 20–30% scored 26% of the time, those rated 40–50% scored 47%.
Games where the model's total differs most from the posted line, with the points actually scored.
On average the model projects 0.4 points a game below the market. Totals use team statistics only, so they do not drop when a quarterback is ruled out.
Ranked best to worst going into this week. 50 is an average team, and each point of rating is worth about a fifth of a point on the scoreboard. Ratings update with this week’s results when the next week is built. Arrows show the move since last week. Tap a team for its page.
Player ratings blend this season with 2025 and 2024. The weight at each position was set by testing how well the first three weeks plus earlier years predicted the rest of 2024 and 2025.
| Position | Weight on past seasons | Error cut, old weights | Error cut, now |
|---|---|---|---|
| RB | 0.25 | about 0% | about 0% |
| WR | 0.50 | about 0% | about 0% |
| TE | 0.25 | about 0% | about 0% |
| EDGE | 1.00 | 22% | 22% |
| DT | 1.00 | 15% | 15% |
| LB | 1.50 | 12% | 16% |
| CB | 0.50 | about 0% | about 0% |
| S | 1.50 | 15% | 16% |
History is predictive for pass rushers, interior linemen, linebackers and safeties. For running backs, receivers, tight ends and corners last year adds almost nothing, so this season's form leads. Team ratings carry over 45% of 2025's margin, weighted like eight games of 2026 data.
Toss-up Picks the model rated under 60% won 55% of 526 held-out games (2021 to 2025).
The model had the Seahawks by about 1 with 45 total points.
Each bar pushes the prediction toward one team. Read down to the total.
Team strength blends this season's results with a share of last season's, then injuries and home field are applied.
Every way the margin can land. The taller the curve, the more likely that result.
47% chance of a one-score game (8 points or fewer). NFL results land about 13 points either side of the predicted margin, so a blowout either way stays on the table.
Tap a player to rule him out or put him back in, and watch the prediction move. Useful when late team news breaks.
Each player's value is the points his team loses if he sits and his usual replacement plays. Offensive players change their own team's score; defenders change the opponent's.
Two independent opinions. Where they split is where to look harder.
| Model | Market | Result | |
|---|---|---|---|
| Broncos chance of winning | 48% | 42% | Won |
| Broncos winning margin | -0.7 | -2.5 | 3 |
| Total points | 45.0 | 40.5 | 31 |
The model is 1.8 points more on the Broncos. Treat a gap as a prompt to check team news, not as an edge: across 3,663 back-tested games since 2012, the side the model favoured against the spread covered 49% of the time, no better than a coin.
Each square is one simulated game, placed by its winning margin. Play one, or a thousand, and watch the model's odds take shape.
How the game is likely to be played, and the conditions around it.
The marker is the league average. Tendencies rest on 5 games and are pulled toward the average. The shift for leading or trailing was measured on the 2025 season. These indicators are for context and do not yet feed the yardage numbers.
Travel is measured from the visiting team's home stadium. None of these factors change the prediction apart from home field.
Where each team ranks among the 32 on its 2026 statistics before this game. A strength is top eight in the league; a weakness is bottom eight.
Ranks rest on 5 games a team, so they can move quickly. They describe what has happened so far, not the model's prediction, which also weighs last season and injuries.
Chance each player scores a rushing or receiving touchdown, and the price that would be fair for it.
If a bookmaker offers more than the fair price, the price is better than the model thinks it should be; most prices will be shorter because of the bookmaker's margin. Tested on 3,072 player-games from 2025: players rated 20–30% scored 26% of the time, those rated 40–50% scored 47%. Players ruled out are excluded and questionable players are flagged.
The bar covers the middle 80% of simulated games; the coloured mark is the midpoint. The dark line is what the player actually did. Wide bars are normal: yardage swings a lot from week to week.
Ranges come from testing these projections on 38,722 player-games in 2024 and 2025. The typical miss was 60 yards passing, 25 rushing and 23 receiving. Set your own over and under lines in the Parlay builder.
Combinations inside this game, priced from 60,000 simulated versions of it so that legs which move together are counted properly.
| Combination | Model chance | Fair price |
|---|---|---|
| ✕ MissedMost likely pairSeahawks to win + Under 40.5 points | 26% 26% if independent | $3.82 |
| ✕ MissedMost likely pairSeahawks to win + Over 40.5 points | 26% 26% if independent | $3.88 |
| ✕ MissedMost likely trioSeahawks to win + Over 40.5 points + Jaxon Smith-Njigba anytime TD | 14% 11% if independent | $6.98 |
| ✕ MissedMost likely trioSeahawks −2.5 + Over 40.5 points + Jaxon Smith-Njigba anytime TD | 14% 10% if independent | $7.25 |
| ✕ MissedLegs that move togetherSeahawks −2.5 + Jaxon Smith-Njigba anytime TD + Emanuel Wilson anytime TD | 11% 7.1% if independent | $9.32 |
| ✕ MissedLegs that move togetherSeahawks to win + Jaxon Smith-Njigba anytime TD + Emanuel Wilson anytime TD | 11% 7.4% if independent | $9.04 |
A win makes the over and big touchdown days more likely, so those legs reinforce each other. A favourite plus the under pull against each other. Bookmakers price same-game multis with their own margin, so what you are offered will be shorter than the fair price.
What each offence is expected to produce against the other defence. The stronger side in each row is in full colour.
Each figure is the league average plus that offence's own edge plus the opposing defence's edge, after pulling small samples toward the league average.
2026 season before this game, per game or per play.
Points per game each player is worth above a replacement, from pass rush, coverage, tackles for loss and production relative to teammates.
Active players only. Offensive linemen have no public statistics, so they are not rated here.
Points of margin lost this week from players on the injury report or injured reserve.
Out and Doubtful count as 0% available; Questionable counts as 50% to 75% depending on Friday practice. Players with no game status are treated as playing.
2026 results before this game, most recent last.
Regular-season games since 2022, most recent first.
Lean Picks the model rated 60 to 70% won 66% of 426 held-out games (2021 to 2025).
Jaguars by about 4, with 41 total points. Played ten times, the Jaguars would win about 6. It is a lean, not a lock.
Each bar pushes the prediction toward one team. Read down to the total.
Team strength blends this season's results with a share of last season's, then injuries and home field are applied.
Every way the margin can land. The taller the curve, the more likely that result.
45% chance of a one-score game (8 points or fewer). NFL results land about 13 points either side of the predicted margin, so a blowout either way stays on the table.
Tap a player to rule him out or put him back in, and watch the prediction move. Useful when late team news breaks.
Each player's value is the points his team loses if he sits and his usual replacement plays. Offensive players change their own team's score; defenders change the opponent's.
Two independent opinions. Where they split is where to look harder.
| Model | Market | |
|---|---|---|
| Jaguars chance of winning | 62% | 57% |
| Jaguars winning margin | 4.0 | 2.5 |
| Total points | 40.6 | 42.5 |
The model is 1.5 points more on the Jaguars. Treat a gap as a prompt to check team news, not as an edge: across 3,663 back-tested games since 2012, the side the model favoured against the spread covered 49% of the time, no better than a coin.
Each square is one simulated game, placed by its winning margin. Play one, or a thousand, and watch the model's odds take shape.
How the game is likely to be played, and the conditions around it.
The marker is the league average. Tendencies rest on 5 games and are pulled toward the average. The shift for leading or trailing was measured on the 2025 season. These indicators are for context and do not yet feed the yardage numbers.
Travel is measured from the visiting team's home stadium. None of these factors change the prediction apart from home field.
Where each team ranks among the 32 on its 2026 statistics before this game. A strength is top eight in the league; a weakness is bottom eight.
Ranks rest on 5 games a team, so they can move quickly. They describe what has happened so far, not the model's prediction, which also weighs last season and injuries.
Chance each player scores a rushing or receiving touchdown, and the price that would be fair for it.
If a bookmaker offers more than the fair price, the price is better than the model thinks it should be; most prices will be shorter because of the bookmaker's margin. Tested on 3,072 player-games from 2025: players rated 20–30% scored 26% of the time, those rated 40–50% scored 47%. Players ruled out are excluded and questionable players are flagged.
The bar covers the middle 80% of simulated games; the coloured mark is the midpoint. Wide bars are normal: yardage swings a lot from week to week.
Ranges come from testing these projections on 38,722 player-games in 2024 and 2025. The typical miss was 60 yards passing, 25 rushing and 23 receiving. Set your own over and under lines in the Parlay builder.
Combinations inside this game, priced from 60,000 simulated versions of it so that legs which move together are counted properly.
| Combination | Model chance | Fair price |
|---|---|---|
| Most likely pairJaguars to win + Over 42.5 points | 31% 31% if independent | $3.23 |
| Most likely pairJaguars to win + Under 42.5 points | 30% 31% if independent | $3.28 |
| Most likely trioJaguars to win + Over 42.5 points + Bhayshul Tuten anytime TD | 14% 11% if independent | $6.99 |
| Most likely trioJaguars −2.5 + Over 42.5 points + Bhayshul Tuten anytime TD | 12% 8.7% if independent | $8.40 |
| Legs that move togetherTexans +2.5 + Over 42.5 points + Nico Collins anytime TD | 11% 7.6% if independent | $9.07 |
| Legs that move togetherJaguars to win + Over 42.5 points + Parker Washington anytime TD | 11% 7.8% if independent | $9.40 |
A win makes the over and big touchdown days more likely, so those legs reinforce each other. A favourite plus the under pull against each other. Bookmakers price same-game multis with their own margin, so what you are offered will be shorter than the fair price.
What each offence is expected to produce against the other defence. The stronger side in each row is in full colour.
Each figure is the league average plus that offence's own edge plus the opposing defence's edge, after pulling small samples toward the league average.
2026 season before this game, per game or per play.
Points per game each player is worth above a replacement, from pass rush, coverage, tackles for loss and production relative to teammates.
Active players only. Offensive linemen have no public statistics, so they are not rated here.
Points of margin lost this week from players on the injury report or injured reserve.
Out and Doubtful count as 0% available; Questionable counts as 50% to 75% depending on Friday practice. Players with no game status are treated as playing.
2026 results before this game, most recent last.
Regular-season games since 2022, most recent first.
Toss-up Picks the model rated under 60% won 55% of 526 held-out games (2021 to 2025).
Falcons by about 1, with 48 total points. Played ten times, the Falcons would win about 5. Less than a field goal separates them, so treat it as a coin flip.
Each bar pushes the prediction toward one team. Read down to the total.
Team strength blends this season's results with a share of last season's, then injuries and home field are applied.
Every way the margin can land. The taller the curve, the more likely that result.
47% chance of a one-score game (8 points or fewer). NFL results land about 13 points either side of the predicted margin, so a blowout either way stays on the table.
Tap a player to rule him out or put him back in, and watch the prediction move. Useful when late team news breaks.
Each player's value is the points his team loses if he sits and his usual replacement plays. Offensive players change their own team's score; defenders change the opponent's.
Two independent opinions. Where they split is where to look harder.
| Model | Market | |
|---|---|---|
| Falcons chance of winning | 53% | 48% |
| Falcons winning margin | 0.9 | -1.5 |
| Total points | 48.0 | 51.5 |
The model is 2.4 points more on the Falcons. Treat a gap as a prompt to check team news, not as an edge: across 3,663 back-tested games since 2012, the side the model favoured against the spread covered 49% of the time, no better than a coin.
Each square is one simulated game, placed by its winning margin. Play one, or a thousand, and watch the model's odds take shape.
How the game is likely to be played, and the conditions around it.
The marker is the league average. Tendencies rest on 5 games and are pulled toward the average. The shift for leading or trailing was measured on the 2025 season. These indicators are for context and do not yet feed the yardage numbers.
Indoors, so wind and rain do not come into it.
Travel is measured from the visiting team's home stadium. None of these factors change the prediction apart from home field.
Where each team ranks among the 32 on its 2026 statistics before this game. A strength is top eight in the league; a weakness is bottom eight.
Ranks rest on 5 games a team, so they can move quickly. They describe what has happened so far, not the model's prediction, which also weighs last season and injuries.
Chance each player scores a rushing or receiving touchdown, and the price that would be fair for it.
If a bookmaker offers more than the fair price, the price is better than the model thinks it should be; most prices will be shorter because of the bookmaker's margin. Tested on 3,072 player-games from 2025: players rated 20–30% scored 26% of the time, those rated 40–50% scored 47%. Players ruled out are excluded and questionable players are flagged.
The bar covers the middle 80% of simulated games; the coloured mark is the midpoint. Wide bars are normal: yardage swings a lot from week to week.
Ranges come from testing these projections on 38,722 player-games in 2024 and 2025. The typical miss was 60 yards passing, 25 rushing and 23 receiving. Set your own over and under lines in the Parlay builder.
Combinations inside this game, priced from 60,000 simulated versions of it so that legs which move together are counted properly.
| Combination | Model chance | Fair price |
|---|---|---|
| Most likely pairFalcons to win + Bijan Robinson anytime TD | 34% 30% if independent | $2.95 |
| Most likely pairFalcons +1.5 + Bijan Robinson anytime TD | 32% 28% if independent | $3.08 |
| Most likely trioFalcons to win + Over 51.5 points + Bijan Robinson anytime TD | 19% 15% if independent | $5.29 |
| Most likely trioFalcons +1.5 + Over 51.5 points + Bijan Robinson anytime TD | 18% 14% if independent | $5.54 |
| Legs that move togetherFalcons +1.5 + Bijan Robinson anytime TD + Drake London anytime TD | 13% 9.4% if independent | $7.56 |
| Legs that move togetherFalcons +1.5 + Bijan Robinson anytime TD + Brian Robinson anytime TD | 12% 8.3% if independent | $8.57 |
A win makes the over and big touchdown days more likely, so those legs reinforce each other. A favourite plus the under pull against each other. Bookmakers price same-game multis with their own margin, so what you are offered will be shorter than the fair price.
What each offence is expected to produce against the other defence. The stronger side in each row is in full colour.
Each figure is the league average plus that offence's own edge plus the opposing defence's edge, after pulling small samples toward the league average.
2026 season before this game, per game or per play.
Points per game each player is worth above a replacement, from pass rush, coverage, tackles for loss and production relative to teammates.
Active players only. Offensive linemen have no public statistics, so they are not rated here.
Points of margin lost this week from players on the injury report or injured reserve.
Out and Doubtful count as 0% available; Questionable counts as 50% to 75% depending on Friday practice. Players with no game status are treated as playing.
2026 results before this game, most recent last.
Regular-season games since 2022, most recent first.
Toss-up Picks the model rated under 60% won 55% of 526 held-out games (2021 to 2025).
Browns by about 2, with 42 total points. Played ten times, the Browns would win about 6. Less than a field goal separates them, so treat it as a coin flip.
Each bar pushes the prediction toward one team. Read down to the total.
Team strength blends this season's results with a share of last season's, then injuries and home field are applied.
Every way the margin can land. The taller the curve, the more likely that result.
47% chance of a one-score game (8 points or fewer). NFL results land about 13 points either side of the predicted margin, so a blowout either way stays on the table.
Tap a player to rule him out or put him back in, and watch the prediction move. Useful when late team news breaks.
Each player's value is the points his team loses if he sits and his usual replacement plays. Offensive players change their own team's score; defenders change the opponent's.
Two independent opinions. Where they split is where to look harder.
| Model | Market | |
|---|---|---|
| Browns chance of winning | 56% | 40% |
| Browns winning margin | 2.1 | -3.0 |
| Total points | 41.8 | 39.5 |
The model is 5.1 points more on the Browns. Treat a gap as a prompt to check team news, not as an edge: across 3,663 back-tested games since 2012, the side the model favoured against the spread covered 49% of the time, no better than a coin.
Each square is one simulated game, placed by its winning margin. Play one, or a thousand, and watch the model's odds take shape.
How the game is likely to be played, and the conditions around it.
The marker is the league average. Tendencies rest on 5 games and are pulled toward the average. The shift for leading or trailing was measured on the 2025 season. These indicators are for context and do not yet feed the yardage numbers.
Travel is measured from the visiting team's home stadium. None of these factors change the prediction apart from home field.
Where each team ranks among the 32 on its 2026 statistics before this game. A strength is top eight in the league; a weakness is bottom eight.
Ranks rest on 5 games a team, so they can move quickly. They describe what has happened so far, not the model's prediction, which also weighs last season and injuries.
Chance each player scores a rushing or receiving touchdown, and the price that would be fair for it.
If a bookmaker offers more than the fair price, the price is better than the model thinks it should be; most prices will be shorter because of the bookmaker's margin. Tested on 3,072 player-games from 2025: players rated 20–30% scored 26% of the time, those rated 40–50% scored 47%. Players ruled out are excluded and questionable players are flagged.
The bar covers the middle 80% of simulated games; the coloured mark is the midpoint. Wide bars are normal: yardage swings a lot from week to week.
Ranges come from testing these projections on 38,722 player-games in 2024 and 2025. The typical miss was 60 yards passing, 25 rushing and 23 receiving. Set your own over and under lines in the Parlay builder.
Combinations inside this game, priced from 60,000 simulated versions of it so that legs which move together are counted properly.
| Combination | Model chance | Fair price |
|---|---|---|
| Most likely pairRavens −3 + Derrick Henry anytime TD | 33% 28% if independent | $3.01 |
| Most likely pairOver 39.5 points + Derrick Henry anytime TD | 33% 28% if independent | $3.05 |
| Most likely trioRavens −3 + Over 39.5 points + Derrick Henry anytime TD | 18% 14% if independent | $5.43 |
| Most likely trioBrowns to win + Over 39.5 points + Derrick Henry anytime TD | 17% 16% if independent | $6.06 |
| Legs that move togetherRavens to win + Derrick Henry anytime TD + Zay Flowers anytime TD | 11% 6.9% if independent | $9.16 |
| Legs that move togetherRavens −3 + Derrick Henry anytime TD + Zay Flowers anytime TD | 12% 7.8% if independent | $8.35 |
A win makes the over and big touchdown days more likely, so those legs reinforce each other. A favourite plus the under pull against each other. Bookmakers price same-game multis with their own margin, so what you are offered will be shorter than the fair price.
What each offence is expected to produce against the other defence. The stronger side in each row is in full colour.
Each figure is the league average plus that offence's own edge plus the opposing defence's edge, after pulling small samples toward the league average.
2026 season before this game, per game or per play.
Points per game each player is worth above a replacement, from pass rush, coverage, tackles for loss and production relative to teammates.
Active players only. Offensive linemen have no public statistics, so they are not rated here.
Points of margin lost this week from players on the injury report or injured reserve.
Out and Doubtful count as 0% available; Questionable counts as 50% to 75% depending on Friday practice. Players with no game status are treated as playing.
2026 results before this game, most recent last.
Regular-season games since 2022, most recent first.
Strong pick Picks the model rated 70 to 80% won 74% of 265 held-out games (2021 to 2025).
Colts by about 9, with 45 total points. Played ten times, the Colts would win about 8. A clear favourite, though NFL games still turn on a handful of plays.
Each bar pushes the prediction toward one team. Read down to the total.
Team strength blends this season's results with a share of last season's, then injuries and home field are applied.
Every way the margin can land. The taller the curve, the more likely that result.
38% chance of a one-score game (8 points or fewer). NFL results land about 13 points either side of the predicted margin, so a blowout either way stays on the table.
Tap a player to rule him out or put him back in, and watch the prediction move. Useful when late team news breaks.
Each player's value is the points his team loses if he sits and his usual replacement plays. Offensive players change their own team's score; defenders change the opponent's.
Two independent opinions. Where they split is where to look harder.
| Model | Market | |
|---|---|---|
| Colts chance of winning | 75% | 74% |
| Colts winning margin | 9.2 | 7.0 |
| Total points | 45.4 | 44.5 |
The model is 2.2 points more on the Colts. Treat a gap as a prompt to check team news, not as an edge: across 3,663 back-tested games since 2012, the side the model favoured against the spread covered 49% of the time, no better than a coin.
Each square is one simulated game, placed by its winning margin. Play one, or a thousand, and watch the model's odds take shape.
How the game is likely to be played, and the conditions around it.
The marker is the league average. Tendencies rest on 5 games and are pulled toward the average. The shift for leading or trailing was measured on the 2025 season. These indicators are for context and do not yet feed the yardage numbers.
Indoors, so wind and rain do not come into it.
Travel is measured from the visiting team's home stadium. None of these factors change the prediction apart from home field.
Where each team ranks among the 32 on its 2026 statistics before this game. A strength is top eight in the league; a weakness is bottom eight.
Ranks rest on 5 games a team, so they can move quickly. They describe what has happened so far, not the model's prediction, which also weighs last season and injuries.
Chance each player scores a rushing or receiving touchdown, and the price that would be fair for it.
If a bookmaker offers more than the fair price, the price is better than the model thinks it should be; most prices will be shorter because of the bookmaker's margin. Tested on 3,072 player-games from 2025: players rated 20–30% scored 26% of the time, those rated 40–50% scored 47%. Players ruled out are excluded and questionable players are flagged.
The bar covers the middle 80% of simulated games; the coloured mark is the midpoint. Wide bars are normal: yardage swings a lot from week to week.
Ranges come from testing these projections on 38,722 player-games in 2024 and 2025. The typical miss was 60 yards passing, 25 rushing and 23 receiving. Set your own over and under lines in the Parlay builder.
Combinations inside this game, priced from 60,000 simulated versions of it so that legs which move together are counted properly.
| Combination | Model chance | Fair price |
|---|---|---|
| Most likely pairColts to win + Jonathan Taylor anytime TD | 52% 49% if independent | $1.92 |
| Most likely pairColts to win + Under 44.5 points | 38% 38% if independent | $2.66 |
| Most likely trioColts to win + Over 44.5 points + Jonathan Taylor anytime TD | 28% 24% if independent | $3.53 |
| Most likely trioColts to win + Under 44.5 points + Jonathan Taylor anytime TD | 24% 24% if independent | $4.21 |
| Legs that move togetherOver 44.5 points + Tony Pollard anytime TD + Jonathan Taylor anytime TD | 11% 8.3% if independent | $8.84 |
| Legs that move togetherColts −7 + Jonathan Taylor anytime TD + Josh Downs anytime TD | 12% 9.1% if independent | $8.14 |
A win makes the over and big touchdown days more likely, so those legs reinforce each other. A favourite plus the under pull against each other. Bookmakers price same-game multis with their own margin, so what you are offered will be shorter than the fair price.
What each offence is expected to produce against the other defence. The stronger side in each row is in full colour.
Each figure is the league average plus that offence's own edge plus the opposing defence's edge, after pulling small samples toward the league average.
2026 season before this game, per game or per play.
Points per game each player is worth above a replacement, from pass rush, coverage, tackles for loss and production relative to teammates.
Active players only. Offensive linemen have no public statistics, so they are not rated here.
Points of margin lost this week from players on the injury report or injured reserve.
Out and Doubtful count as 0% available; Questionable counts as 50% to 75% depending on Friday practice. Players with no game status are treated as playing.
2026 results before this game, most recent last.
Regular-season games since 2022, most recent first.
Very strong Picks the model rated 80% or more won 80% of 138 held-out games (2021 to 2025).
Patriots by about 12, with 44 total points. Played ten times, the Patriots would win about 8. A clear favourite, though NFL games still turn on a handful of plays.
Each bar pushes the prediction toward one team. Read down to the total.
Team strength blends this season's results with a share of last season's, then injuries and home field are applied.
Every way the margin can land. The taller the curve, the more likely that result.
34% chance of a one-score game (8 points or fewer). NFL results land about 13 points either side of the predicted margin, so a blowout either way stays on the table.
Tap a player to rule him out or put him back in, and watch the prediction move. Useful when late team news breaks.
Each player's value is the points his team loses if he sits and his usual replacement plays. Offensive players change their own team's score; defenders change the opponent's.
Two independent opinions. Where they split is where to look harder.
| Model | Market | |
|---|---|---|
| Patriots chance of winning | 81% | 74% |
| Patriots winning margin | 11.6 | 7.5 |
| Total points | 44.0 | 43.5 |
The model is 4.1 points more on the Patriots. Treat a gap as a prompt to check team news, not as an edge: across 3,663 back-tested games since 2012, the side the model favoured against the spread covered 49% of the time, no better than a coin.
Each square is one simulated game, placed by its winning margin. Play one, or a thousand, and watch the model's odds take shape.
How the game is likely to be played, and the conditions around it.
The marker is the league average. Tendencies rest on 5 games and are pulled toward the average. The shift for leading or trailing was measured on the 2025 season. These indicators are for context and do not yet feed the yardage numbers.
Travel is measured from the visiting team's home stadium. None of these factors change the prediction apart from home field.
Where each team ranks among the 32 on its 2026 statistics before this game. A strength is top eight in the league; a weakness is bottom eight.
Ranks rest on 5 games a team, so they can move quickly. They describe what has happened so far, not the model's prediction, which also weighs last season and injuries.
Chance each player scores a rushing or receiving touchdown, and the price that would be fair for it.
If a bookmaker offers more than the fair price, the price is better than the model thinks it should be; most prices will be shorter because of the bookmaker's margin. Tested on 3,072 player-games from 2025: players rated 20–30% scored 26% of the time, those rated 40–50% scored 47%. Players ruled out are excluded and questionable players are flagged.
The bar covers the middle 80% of simulated games; the coloured mark is the midpoint. Wide bars are normal: yardage swings a lot from week to week.
Ranges come from testing these projections on 38,722 player-games in 2024 and 2025. The typical miss was 60 yards passing, 25 rushing and 23 receiving. Set your own over and under lines in the Parlay builder.
Combinations inside this game, priced from 60,000 simulated versions of it so that legs which move together are counted properly.
| Combination | Model chance | Fair price |
|---|---|---|
| Most likely pairPatriots to win + Under 43.5 points | 40% 41% if independent | $2.47 |
| Most likely pairPatriots to win + Over 43.5 points | 40% 40% if independent | $2.48 |
| Most likely trioPatriots to win + Over 43.5 points + Rhamondre Stevenson anytime TD | 20% 17% if independent | $4.94 |
| Most likely trioPatriots to win + Under 43.5 points + Rhamondre Stevenson anytime TD | 16% 17% if independent | $6.32 |
| Legs that move togetherJets +7.5 + Over 43.5 points + Braelon Allen anytime TD | 10% 6.8% if independent | $9.60 |
| Legs that move togetherJets +7.5 + Over 43.5 points + Garrett Wilson anytime TD | 11% 7.1% if independent | $9.28 |
A win makes the over and big touchdown days more likely, so those legs reinforce each other. A favourite plus the under pull against each other. Bookmakers price same-game multis with their own margin, so what you are offered will be shorter than the fair price.
What each offence is expected to produce against the other defence. The stronger side in each row is in full colour.
Each figure is the league average plus that offence's own edge plus the opposing defence's edge, after pulling small samples toward the league average.
2026 season before this game, per game or per play.
Points per game each player is worth above a replacement, from pass rush, coverage, tackles for loss and production relative to teammates.
Active players only. Offensive linemen have no public statistics, so they are not rated here.
Points of margin lost this week from players on the injury report or injured reserve.
Out and Doubtful count as 0% available; Questionable counts as 50% to 75% depending on Friday practice. Players with no game status are treated as playing.
2026 results before this game, most recent last.
Regular-season games since 2022, most recent first.
Toss-up Picks the model rated under 60% won 55% of 526 held-out games (2021 to 2025).
Giants by about 2, with 44 total points. Played ten times, the Giants would win about 6. Less than a field goal separates them, so treat it as a coin flip.
Each bar pushes the prediction toward one team. Read down to the total.
Team strength blends this season's results with a share of last season's, then injuries and home field are applied.
Every way the margin can land. The taller the curve, the more likely that result.
47% chance of a one-score game (8 points or fewer). NFL results land about 13 points either side of the predicted margin, so a blowout either way stays on the table.
Tap a player to rule him out or put him back in, and watch the prediction move. Useful when late team news breaks.
Each player's value is the points his team loses if he sits and his usual replacement plays. Offensive players change their own team's score; defenders change the opponent's.
Two independent opinions. Where they split is where to look harder.
| Model | Market | |
|---|---|---|
| Giants chance of winning | 55% | 50% |
| Giants winning margin | 1.7 | -1.5 |
| Total points | 43.5 | 46.5 |
The model is 3.2 points more on the Giants. Treat a gap as a prompt to check team news, not as an edge: across 3,663 back-tested games since 2012, the side the model favoured against the spread covered 49% of the time, no better than a coin.
Each square is one simulated game, placed by its winning margin. Play one, or a thousand, and watch the model's odds take shape.
How the game is likely to be played, and the conditions around it.
The marker is the league average. Tendencies rest on 5 games and are pulled toward the average. The shift for leading or trailing was measured on the 2025 season. These indicators are for context and do not yet feed the yardage numbers.
Travel is measured from the visiting team's home stadium. None of these factors change the prediction apart from home field.
Where each team ranks among the 32 on its 2026 statistics before this game. A strength is top eight in the league; a weakness is bottom eight.
Ranks rest on 5 games a team, so they can move quickly. They describe what has happened so far, not the model's prediction, which also weighs last season and injuries.
Chance each player scores a rushing or receiving touchdown, and the price that would be fair for it.
If a bookmaker offers more than the fair price, the price is better than the model thinks it should be; most prices will be shorter because of the bookmaker's margin. Tested on 3,072 player-games from 2025: players rated 20–30% scored 26% of the time, those rated 40–50% scored 47%. Players ruled out are excluded and questionable players are flagged.
The bar covers the middle 80% of simulated games; the coloured mark is the midpoint. Wide bars are normal: yardage swings a lot from week to week.
Ranges come from testing these projections on 38,722 player-games in 2024 and 2025. The typical miss was 60 yards passing, 25 rushing and 23 receiving. Set your own over and under lines in the Parlay builder.
Combinations inside this game, priced from 60,000 simulated versions of it so that legs which move together are counted properly.
| Combination | Model chance | Fair price |
|---|---|---|
| Most likely pairGiants to win + Over 46.5 points | 28% 27% if independent | $3.62 |
| Most likely pairGiants to win + Under 46.5 points | 28% 28% if independent | $3.62 |
| Most likely trioGiants to win + Over 46.5 points + Cam Skattebo anytime TD | 15% 11% if independent | $6.84 |
| Most likely trioGiants to win + Over 46.5 points + Malik Nabers anytime TD | 14% 10% if independent | $7.21 |
| Legs that move togetherOver 46.5 points + Cam Skattebo anytime TD + Malik Nabers anytime TD | 11% 7.3% if independent | $9.31 |
| Legs that move togetherGiants +1.5 + Cam Skattebo anytime TD + Malik Nabers anytime TD | 11% 7.4% if independent | $9.22 |
A win makes the over and big touchdown days more likely, so those legs reinforce each other. A favourite plus the under pull against each other. Bookmakers price same-game multis with their own margin, so what you are offered will be shorter than the fair price.
What each offence is expected to produce against the other defence. The stronger side in each row is in full colour.
Each figure is the league average plus that offence's own edge plus the opposing defence's edge, after pulling small samples toward the league average.
2026 season before this game, per game or per play.
Points per game each player is worth above a replacement, from pass rush, coverage, tackles for loss and production relative to teammates.
Active players only. Offensive linemen have no public statistics, so they are not rated here.
Points of margin lost this week from players on the injury report or injured reserve.
Out and Doubtful count as 0% available; Questionable counts as 50% to 75% depending on Friday practice. Players with no game status are treated as playing.
2026 results before this game, most recent last.
Regular-season games since 2022, most recent first.
Lean Picks the model rated 60 to 70% won 66% of 426 held-out games (2021 to 2025).
Panthers by about 6, with 42 total points. Played ten times, the Panthers would win about 7. It is a lean, not a lock.
Each bar pushes the prediction toward one team. Read down to the total.
Team strength blends this season's results with a share of last season's, then injuries and home field are applied.
Every way the margin can land. The taller the curve, the more likely that result.
43% chance of a one-score game (8 points or fewer). NFL results land about 13 points either side of the predicted margin, so a blowout either way stays on the table.
Tap a player to rule him out or put him back in, and watch the prediction move. Useful when late team news breaks.
Each player's value is the points his team loses if he sits and his usual replacement plays. Offensive players change their own team's score; defenders change the opponent's.
Two independent opinions. Where they split is where to look harder.
| Model | Market | |
|---|---|---|
| Eagles chance of winning | 33% | 40% |
| Eagles winning margin | -5.9 | -3.5 |
| Total points | 42.0 | 39.5 |
The model is 2.4 points more on the Panthers. Treat a gap as a prompt to check team news, not as an edge: across 3,663 back-tested games since 2012, the side the model favoured against the spread covered 49% of the time, no better than a coin.
Each square is one simulated game, placed by its winning margin. Play one, or a thousand, and watch the model's odds take shape.
How the game is likely to be played, and the conditions around it.
The marker is the league average. Tendencies rest on 5 games and are pulled toward the average. The shift for leading or trailing was measured on the 2025 season. These indicators are for context and do not yet feed the yardage numbers.
Travel is measured from the visiting team's home stadium. None of these factors change the prediction apart from home field.
Where each team ranks among the 32 on its 2026 statistics before this game. A strength is top eight in the league; a weakness is bottom eight.
Ranks rest on 5 games a team, so they can move quickly. They describe what has happened so far, not the model's prediction, which also weighs last season and injuries.
Chance each player scores a rushing or receiving touchdown, and the price that would be fair for it.
If a bookmaker offers more than the fair price, the price is better than the model thinks it should be; most prices will be shorter because of the bookmaker's margin. Tested on 3,072 player-games from 2025: players rated 20–30% scored 26% of the time, those rated 40–50% scored 47%. Players ruled out are excluded and questionable players are flagged.
The bar covers the middle 80% of simulated games; the coloured mark is the midpoint. Wide bars are normal: yardage swings a lot from week to week.
Ranges come from testing these projections on 38,722 player-games in 2024 and 2025. The typical miss was 60 yards passing, 25 rushing and 23 receiving. Set your own over and under lines in the Parlay builder.
Combinations inside this game, priced from 60,000 simulated versions of it so that legs which move together are counted properly.
| Combination | Model chance | Fair price |
|---|---|---|
| Most likely pairPanthers to win + Chuba Hubbard anytime TD | 35% 31% if independent | $2.86 |
| Most likely pairPanthers to win + Under 39.5 points | 34% 34% if independent | $2.98 |
| Most likely trioPanthers to win + Over 39.5 points + Chuba Hubbard anytime TD | 20% 16% if independent | $5.08 |
| Most likely trioPanthers to win + Under 39.5 points + Chuba Hubbard anytime TD | 15% 16% if independent | $6.52 |
| Legs that move togetherPanthers −3.5 + Chuba Hubbard anytime TD + Tetairoa McMillan anytime TD | 12% 8.1% if independent | $8.67 |
| Legs that move togetherOver 39.5 points + Chuba Hubbard anytime TD + Tetairoa McMillan anytime TD | 11% 8.0% if independent | $8.80 |
A win makes the over and big touchdown days more likely, so those legs reinforce each other. A favourite plus the under pull against each other. Bookmakers price same-game multis with their own margin, so what you are offered will be shorter than the fair price.
What each offence is expected to produce against the other defence. The stronger side in each row is in full colour.
Each figure is the league average plus that offence's own edge plus the opposing defence's edge, after pulling small samples toward the league average.
2026 season before this game, per game or per play.
Points per game each player is worth above a replacement, from pass rush, coverage, tackles for loss and production relative to teammates.
Active players only. Offensive linemen have no public statistics, so they are not rated here.
Points of margin lost this week from players on the injury report or injured reserve.
Out and Doubtful count as 0% available; Questionable counts as 50% to 75% depending on Friday practice. Players with no game status are treated as playing.
2026 results before this game, most recent last.
Regular-season games since 2022, most recent first.
Toss-up Picks the model rated under 60% won 55% of 526 held-out games (2021 to 2025).
Steelers by about 1, with 41 total points. Played ten times, the Steelers would win about 5. Less than a field goal separates them, so treat it as a coin flip.
Each bar pushes the prediction toward one team. Read down to the total.
Team strength blends this season's results with a share of last season's, then injuries and home field are applied.
Every way the margin can land. The taller the curve, the more likely that result.
47% chance of a one-score game (8 points or fewer). NFL results land about 13 points either side of the predicted margin, so a blowout either way stays on the table.
Tap a player to rule him out or put him back in, and watch the prediction move. Useful when late team news breaks.
Each player's value is the points his team loses if he sits and his usual replacement plays. Offensive players change their own team's score; defenders change the opponent's.
Two independent opinions. Where they split is where to look harder.
| Model | Market | |
|---|---|---|
| Buccaneers chance of winning | 47% | 43% |
| Buccaneers winning margin | -1.1 | -2.5 |
| Total points | 40.9 | 39.5 |
The model is 1.4 points more on the Buccaneers. Treat a gap as a prompt to check team news, not as an edge: across 3,663 back-tested games since 2012, the side the model favoured against the spread covered 49% of the time, no better than a coin.
Each square is one simulated game, placed by its winning margin. Play one, or a thousand, and watch the model's odds take shape.
How the game is likely to be played, and the conditions around it.
The marker is the league average. Tendencies rest on 5 games and are pulled toward the average. The shift for leading or trailing was measured on the 2025 season. These indicators are for context and do not yet feed the yardage numbers.
Travel is measured from the visiting team's home stadium. None of these factors change the prediction apart from home field.
Where each team ranks among the 32 on its 2026 statistics before this game. A strength is top eight in the league; a weakness is bottom eight.
Ranks rest on 5 games a team, so they can move quickly. They describe what has happened so far, not the model's prediction, which also weighs last season and injuries.
Chance each player scores a rushing or receiving touchdown, and the price that would be fair for it.
If a bookmaker offers more than the fair price, the price is better than the model thinks it should be; most prices will be shorter because of the bookmaker's margin. Tested on 3,072 player-games from 2025: players rated 20–30% scored 26% of the time, those rated 40–50% scored 47%. Players ruled out are excluded and questionable players are flagged.
The bar covers the middle 80% of simulated games; the coloured mark is the midpoint. Wide bars are normal: yardage swings a lot from week to week.
Ranges come from testing these projections on 38,722 player-games in 2024 and 2025. The typical miss was 60 yards passing, 25 rushing and 23 receiving. Set your own over and under lines in the Parlay builder.
Combinations inside this game, priced from 60,000 simulated versions of it so that legs which move together are counted properly.
| Combination | Model chance | Fair price |
|---|---|---|
| Most likely pairSteelers to win + Under 39.5 points | 27% 27% if independent | $3.72 |
| Most likely pairSteelers to win + Over 39.5 points | 27% 27% if independent | $3.75 |
| Most likely trioBuccaneers +2.5 + Over 39.5 points + Bucky Irving anytime TD | 13% 9.1% if independent | $7.84 |
| Most likely trioSteelers to win + Over 39.5 points + Jaylen Warren anytime TD | 12% 8.9% if independent | $8.08 |
| Legs that move togetherBuccaneers to win + Over 39.5 points + Bucky Irving anytime TD | 12% 8.5% if independent | $8.32 |
| Legs that move togetherSteelers −2.5 + Over 39.5 points + Jaylen Warren anytime TD | 12% 8.3% if independent | $8.52 |
A win makes the over and big touchdown days more likely, so those legs reinforce each other. A favourite plus the under pull against each other. Bookmakers price same-game multis with their own margin, so what you are offered will be shorter than the fair price.
What each offence is expected to produce against the other defence. The stronger side in each row is in full colour.
Each figure is the league average plus that offence's own edge plus the opposing defence's edge, after pulling small samples toward the league average.
2026 season before this game, per game or per play.
Points per game each player is worth above a replacement, from pass rush, coverage, tackles for loss and production relative to teammates.
Active players only. Offensive linemen have no public statistics, so they are not rated here.
Points of margin lost this week from players on the injury report or injured reserve.
Out and Doubtful count as 0% available; Questionable counts as 50% to 75% depending on Friday practice. Players with no game status are treated as playing.
2026 results before this game, most recent last.
Regular-season games since 2022, most recent first.
Very strong Picks the model rated 80% or more won 80% of 138 held-out games (2021 to 2025).
Rams by about 14, with 52 total points. Played ten times, the Rams would win about 8. A clear favourite, though NFL games still turn on a handful of plays.
Each bar pushes the prediction toward one team. Read down to the total.
Team strength blends this season's results with a share of last season's, then injuries and home field are applied.
Every way the margin can land. The taller the curve, the more likely that result.
30% chance of a one-score game (8 points or fewer). NFL results land about 13 points either side of the predicted margin, so a blowout either way stays on the table.
Tap a player to rule him out or put him back in, and watch the prediction move. Useful when late team news breaks.
Each player's value is the points his team loses if he sits and his usual replacement plays. Offensive players change their own team's score; defenders change the opponent's.
Two independent opinions. Where they split is where to look harder.
| Model | Market | |
|---|---|---|
| Rams chance of winning | 85% | 85% |
| Rams winning margin | 13.9 | 12.5 |
| Total points | 51.6 | 51.5 |
The model is 1.4 points more on the Rams. Treat a gap as a prompt to check team news, not as an edge: across 3,663 back-tested games since 2012, the side the model favoured against the spread covered 49% of the time, no better than a coin.
Each square is one simulated game, placed by its winning margin. Play one, or a thousand, and watch the model's odds take shape.
How the game is likely to be played, and the conditions around it.
The marker is the league average. Tendencies rest on 5 games and are pulled toward the average. The shift for leading or trailing was measured on the 2025 season. These indicators are for context and do not yet feed the yardage numbers.
Indoors, so wind and rain do not come into it.
Travel is measured from the visiting team's home stadium. None of these factors change the prediction apart from home field.
Where each team ranks among the 32 on its 2026 statistics before this game. A strength is top eight in the league; a weakness is bottom eight.
Ranks rest on 5 games a team, so they can move quickly. They describe what has happened so far, not the model's prediction, which also weighs last season and injuries.
Chance each player scores a rushing or receiving touchdown, and the price that would be fair for it.
If a bookmaker offers more than the fair price, the price is better than the model thinks it should be; most prices will be shorter because of the bookmaker's margin. Tested on 3,072 player-games from 2025: players rated 20–30% scored 26% of the time, those rated 40–50% scored 47%. Players ruled out are excluded and questionable players are flagged.
The bar covers the middle 80% of simulated games; the coloured mark is the midpoint. Wide bars are normal: yardage swings a lot from week to week.
Ranges come from testing these projections on 38,722 player-games in 2024 and 2025. The typical miss was 60 yards passing, 25 rushing and 23 receiving. Set your own over and under lines in the Parlay builder.
Combinations inside this game, priced from 60,000 simulated versions of it so that legs which move together are counted properly.
| Combination | Model chance | Fair price |
|---|---|---|
| Most likely pairRams to win + Kyren Williams anytime TD | 58% 56% if independent | $1.73 |
| Most likely pairRams to win + Puka Nacua anytime TD | 55% 53% if independent | $1.83 |
| Most likely trioRams to win + Kyren Williams anytime TD + Puka Nacua anytime TD | 38% 34% if independent | $2.64 |
| Most likely trioRams to win + Kyren Williams anytime TD + Davante Adams anytime TD | 35% 32% if independent | $2.87 |
| Legs that move togetherCardinals +12.5 + Over 51.5 points + Michael Wilson anytime TD | 10% 7.0% if independent | $9.72 |
| Legs that move togetherCardinals +12.5 + Over 51.5 points + Trey McBride anytime TD | 13% 8.7% if independent | $7.97 |
A win makes the over and big touchdown days more likely, so those legs reinforce each other. A favourite plus the under pull against each other. Bookmakers price same-game multis with their own margin, so what you are offered will be shorter than the fair price.
What each offence is expected to produce against the other defence. The stronger side in each row is in full colour.
Each figure is the league average plus that offence's own edge plus the opposing defence's edge, after pulling small samples toward the league average.
2026 season before this game, per game or per play.
Points per game each player is worth above a replacement, from pass rush, coverage, tackles for loss and production relative to teammates.
Active players only. Offensive linemen have no public statistics, so they are not rated here.
Points of margin lost this week from players on the injury report or injured reserve.
Out and Doubtful count as 0% available; Questionable counts as 50% to 75% depending on Friday practice. Players with no game status are treated as playing.
2026 results before this game, most recent last.
Regular-season games since 2022, most recent first.
Strong pick Picks the model rated 70 to 80% won 74% of 265 held-out games (2021 to 2025).
Chiefs by about 8, with 42 total points. Played ten times, the Chiefs would win about 7. A clear favourite, though NFL games still turn on a handful of plays.
Each bar pushes the prediction toward one team. Read down to the total.
Team strength blends this season's results with a share of last season's, then injuries and home field are applied.
Every way the margin can land. The taller the curve, the more likely that result.
41% chance of a one-score game (8 points or fewer). NFL results land about 13 points either side of the predicted margin, so a blowout either way stays on the table.
Tap a player to rule him out or put him back in, and watch the prediction move. Useful when late team news breaks.
Each player's value is the points his team loses if he sits and his usual replacement plays. Offensive players change their own team's score; defenders change the opponent's.
Two independent opinions. Where they split is where to look harder.
| Model | Market | |
|---|---|---|
| Chiefs chance of winning | 72% | 78% |
| Chiefs winning margin | 7.9 | 7.5 |
| Total points | 42.0 | 41.5 |
The model is in line with the market. Treat a gap as a prompt to check team news, not as an edge: across 3,663 back-tested games since 2012, the side the model favoured against the spread covered 49% of the time, no better than a coin.
Each square is one simulated game, placed by its winning margin. Play one, or a thousand, and watch the model's odds take shape.
How the game is likely to be played, and the conditions around it.
The marker is the league average. Tendencies rest on 5 games and are pulled toward the average. The shift for leading or trailing was measured on the 2025 season. These indicators are for context and do not yet feed the yardage numbers.
Travel is measured from the visiting team's home stadium. None of these factors change the prediction apart from home field.
Where each team ranks among the 32 on its 2026 statistics before this game. A strength is top eight in the league; a weakness is bottom eight.
Ranks rest on 5 games a team, so they can move quickly. They describe what has happened so far, not the model's prediction, which also weighs last season and injuries.
Chance each player scores a rushing or receiving touchdown, and the price that would be fair for it.
If a bookmaker offers more than the fair price, the price is better than the model thinks it should be; most prices will be shorter because of the bookmaker's margin. Tested on 3,072 player-games from 2025: players rated 20–30% scored 26% of the time, those rated 40–50% scored 47%. Players ruled out are excluded and questionable players are flagged.
The bar covers the middle 80% of simulated games; the coloured mark is the midpoint. Wide bars are normal: yardage swings a lot from week to week.
Ranges come from testing these projections on 38,722 player-games in 2024 and 2025. The typical miss was 60 yards passing, 25 rushing and 23 receiving. Set your own over and under lines in the Parlay builder.
Combinations inside this game, priced from 60,000 simulated versions of it so that legs which move together are counted properly.
| Combination | Model chance | Fair price |
|---|---|---|
| Most likely pairChiefs to win + Kenneth Walker III anytime TD | 41% 37% if independent | $2.45 |
| Most likely pairChiefs to win + Over 41.5 points | 36% 36% if independent | $2.77 |
| Most likely trioChiefs to win + Over 41.5 points + Kenneth Walker III anytime TD | 23% 19% if independent | $4.36 |
| Most likely trioChiefs to win + Under 41.5 points + Kenneth Walker III anytime TD | 18% 19% if independent | $5.61 |
| Legs that move togetherChargers +7.5 + Over 41.5 points + Omarion Hampton anytime TD | 11% 7.5% if independent | $8.96 |
| Legs that move togetherChiefs −7.5 + Kenneth Walker III anytime TD + Travis Kelce anytime TD | 11% 8.0% if independent | $8.83 |
A win makes the over and big touchdown days more likely, so those legs reinforce each other. A favourite plus the under pull against each other. Bookmakers price same-game multis with their own margin, so what you are offered will be shorter than the fair price.
What each offence is expected to produce against the other defence. The stronger side in each row is in full colour.
Each figure is the league average plus that offence's own edge plus the opposing defence's edge, after pulling small samples toward the league average.
2026 season before this game, per game or per play.
Points per game each player is worth above a replacement, from pass rush, coverage, tackles for loss and production relative to teammates.
Active players only. Offensive linemen have no public statistics, so they are not rated here.
Points of margin lost this week from players on the injury report or injured reserve.
Out and Doubtful count as 0% available; Questionable counts as 50% to 75% depending on Friday practice. Players with no game status are treated as playing.
2026 results before this game, most recent last.
Regular-season games since 2022, most recent first.
Lean Picks the model rated 60 to 70% won 66% of 426 held-out games (2021 to 2025).
Bills by about 6, with 52 total points. Played ten times, the Bills would win about 7. It is a lean, not a lock.
Each bar pushes the prediction toward one team. Read down to the total.
Team strength blends this season's results with a share of last season's, then injuries and home field are applied.
Every way the margin can land. The taller the curve, the more likely that result.
43% chance of a one-score game (8 points or fewer). NFL results land about 13 points either side of the predicted margin, so a blowout either way stays on the table.
Tap a player to rule him out or put him back in, and watch the prediction move. Useful when late team news breaks.
Each player's value is the points his team loses if he sits and his usual replacement plays. Offensive players change their own team's score; defenders change the opponent's.
Two independent opinions. Where they split is where to look harder.
| Model | Market | |
|---|---|---|
| Raiders chance of winning | 32% | 33% |
| Raiders winning margin | -6.4 | -4.5 |
| Total points | 52.4 | 53.5 |
The model is 1.9 points more on the Bills. Treat a gap as a prompt to check team news, not as an edge: across 3,663 back-tested games since 2012, the side the model favoured against the spread covered 49% of the time, no better than a coin.
Each square is one simulated game, placed by its winning margin. Play one, or a thousand, and watch the model's odds take shape.
How the game is likely to be played, and the conditions around it.
The marker is the league average. Tendencies rest on 5 games and are pulled toward the average. The shift for leading or trailing was measured on the 2025 season. These indicators are for context and do not yet feed the yardage numbers.
Indoors, so wind and rain do not come into it.
Travel is measured from the visiting team's home stadium. None of these factors change the prediction apart from home field.
Where each team ranks among the 32 on its 2026 statistics before this game. A strength is top eight in the league; a weakness is bottom eight.
Ranks rest on 5 games a team, so they can move quickly. They describe what has happened so far, not the model's prediction, which also weighs last season and injuries.
Chance each player scores a rushing or receiving touchdown, and the price that would be fair for it.
If a bookmaker offers more than the fair price, the price is better than the model thinks it should be; most prices will be shorter because of the bookmaker's margin. Tested on 3,072 player-games from 2025: players rated 20–30% scored 26% of the time, those rated 40–50% scored 47%. Players ruled out are excluded and questionable players are flagged.
The bar covers the middle 80% of simulated games; the coloured mark is the midpoint. Wide bars are normal: yardage swings a lot from week to week.
Ranges come from testing these projections on 38,722 player-games in 2024 and 2025. The typical miss was 60 yards passing, 25 rushing and 23 receiving. Set your own over and under lines in the Parlay builder.
Combinations inside this game, priced from 60,000 simulated versions of it so that legs which move together are counted properly.
| Combination | Model chance | Fair price |
|---|---|---|
| Most likely pairBills to win + James Cook anytime TD | 46% 43% if independent | $2.15 |
| Most likely pairBills to win + Josh Allen anytime TD | 38% 35% if independent | $2.61 |
| Most likely trioBills to win + James Cook anytime TD + Josh Allen anytime TD | 27% 22% if independent | $3.75 |
| Most likely trioBills to win + Over 53.5 points + James Cook anytime TD | 25% 22% if independent | $3.93 |
| Legs that move togetherOver 53.5 points + Brock Bowers anytime TD + Ashton Jeanty anytime TD | 12% 8.2% if independent | $8.38 |
| Legs that move togetherRaiders +4.5 + Brock Bowers anytime TD + Ashton Jeanty anytime TD | 12% 8.2% if independent | $8.47 |
A win makes the over and big touchdown days more likely, so those legs reinforce each other. A favourite plus the under pull against each other. Bookmakers price same-game multis with their own margin, so what you are offered will be shorter than the fair price.
What each offence is expected to produce against the other defence. The stronger side in each row is in full colour.
Each figure is the league average plus that offence's own edge plus the opposing defence's edge, after pulling small samples toward the league average.
2026 season before this game, per game or per play.
Points per game each player is worth above a replacement, from pass rush, coverage, tackles for loss and production relative to teammates.
Active players only. Offensive linemen have no public statistics, so they are not rated here.
Points of margin lost this week from players on the injury report or injured reserve.
Out and Doubtful count as 0% available; Questionable counts as 50% to 75% depending on Friday practice. Players with no game status are treated as playing.
2026 results before this game, most recent last.
Regular-season games since 2022, most recent first.
Lean Picks the model rated 60 to 70% won 66% of 426 held-out games (2021 to 2025).
Packers by about 4, with 50 total points. Played ten times, the Packers would win about 6. It is a lean, not a lock.
Each bar pushes the prediction toward one team. Read down to the total.
Team strength blends this season's results with a share of last season's, then injuries and home field are applied.
Every way the margin can land. The taller the curve, the more likely that result.
45% chance of a one-score game (8 points or fewer). NFL results land about 13 points either side of the predicted margin, so a blowout either way stays on the table.
Tap a player to rule him out or put him back in, and watch the prediction move. Useful when late team news breaks.
Each player's value is the points his team loses if he sits and his usual replacement plays. Offensive players change their own team's score; defenders change the opponent's.
Two independent opinions. Where they split is where to look harder.
| Model | Market | |
|---|---|---|
| Packers chance of winning | 62% | 51% |
| Packers winning margin | 4.0 | 1.5 |
| Total points | 50.1 | 53.5 |
The model is 2.5 points more on the Packers. Treat a gap as a prompt to check team news, not as an edge: across 3,663 back-tested games since 2012, the side the model favoured against the spread covered 49% of the time, no better than a coin.
Each square is one simulated game, placed by its winning margin. Play one, or a thousand, and watch the model's odds take shape.
How the game is likely to be played, and the conditions around it.
The marker is the league average. Tendencies rest on 5 games and are pulled toward the average. The shift for leading or trailing was measured on the 2025 season. These indicators are for context and do not yet feed the yardage numbers.
Travel is measured from the visiting team's home stadium. None of these factors change the prediction apart from home field.
Where each team ranks among the 32 on its 2026 statistics before this game. A strength is top eight in the league; a weakness is bottom eight.
Ranks rest on 5 games a team, so they can move quickly. They describe what has happened so far, not the model's prediction, which also weighs last season and injuries.
Chance each player scores a rushing or receiving touchdown, and the price that would be fair for it.
If a bookmaker offers more than the fair price, the price is better than the model thinks it should be; most prices will be shorter because of the bookmaker's margin. Tested on 3,072 player-games from 2025: players rated 20–30% scored 26% of the time, those rated 40–50% scored 47%. Players ruled out are excluded and questionable players are flagged.
The bar covers the middle 80% of simulated games; the coloured mark is the midpoint. Wide bars are normal: yardage swings a lot from week to week.
Ranges come from testing these projections on 38,722 player-games in 2024 and 2025. The typical miss was 60 yards passing, 25 rushing and 23 receiving. Set your own over and under lines in the Parlay builder.
Combinations inside this game, priced from 60,000 simulated versions of it so that legs which move together are counted properly.
| Combination | Model chance | Fair price |
|---|---|---|
| Most likely pairPackers to win + Over 53.5 points | 31% 31% if independent | $3.25 |
| Most likely pairPackers to win + Under 53.5 points | 31% 31% if independent | $3.27 |
| Most likely trioCowboys +1.5 + Over 53.5 points + Javonte Williams anytime TD | 16% 12% if independent | $6.19 |
| Most likely trioPackers to win + Over 53.5 points + Javonte Williams anytime TD | 16% 15% if independent | $6.29 |
| Legs that move togetherCowboys to win + Javonte Williams anytime TD + CeeDee Lamb anytime TD | 11% 7.0% if independent | $9.51 |
| Legs that move togetherCowboys +1.5 + Javonte Williams anytime TD + George Pickens anytime TD | 11% 7.5% if independent | $9.29 |
A win makes the over and big touchdown days more likely, so those legs reinforce each other. A favourite plus the under pull against each other. Bookmakers price same-game multis with their own margin, so what you are offered will be shorter than the fair price.
What each offence is expected to produce against the other defence. The stronger side in each row is in full colour.
Each figure is the league average plus that offence's own edge plus the opposing defence's edge, after pulling small samples toward the league average.
2026 season before this game, per game or per play.
Points per game each player is worth above a replacement, from pass rush, coverage, tackles for loss and production relative to teammates.
Active players only. Offensive linemen have no public statistics, so they are not rated here.
Points of margin lost this week from players on the injury report or injured reserve.
Out and Doubtful count as 0% available; Questionable counts as 50% to 75% depending on Friday practice. Players with no game status are treated as playing.
2026 results before this game, most recent last.
Regular-season games since 2022, most recent first.
Strong pick Picks the model rated 70 to 80% won 74% of 265 held-out games (2021 to 2025).
49ers by about 11, with 47 total points. Played ten times, the 49ers would win about 8. A clear favourite, though NFL games still turn on a handful of plays.
Each bar pushes the prediction toward one team. Read down to the total.
Team strength blends this season's results with a share of last season's, then injuries and home field are applied.
Every way the margin can land. The taller the curve, the more likely that result.
36% chance of a one-score game (8 points or fewer). NFL results land about 13 points either side of the predicted margin, so a blowout either way stays on the table.
Tap a player to rule him out or put him back in, and watch the prediction move. Useful when late team news breaks.
Each player's value is the points his team loses if he sits and his usual replacement plays. Offensive players change their own team's score; defenders change the opponent's.
Two independent opinions. Where they split is where to look harder.
| Model | Market | |
|---|---|---|
| 49ers chance of winning | 79% | 78% |
| 49ers winning margin | 10.8 | 7.5 |
| Total points | 46.6 | 52.5 |
The model is 3.3 points more on the 49ers. Treat a gap as a prompt to check team news, not as an edge: across 3,663 back-tested games since 2012, the side the model favoured against the spread covered 49% of the time, no better than a coin.
Each square is one simulated game, placed by its winning margin. Play one, or a thousand, and watch the model's odds take shape.
How the game is likely to be played, and the conditions around it.
The marker is the league average. Tendencies rest on 5 games and are pulled toward the average. The shift for leading or trailing was measured on the 2025 season. These indicators are for context and do not yet feed the yardage numbers.
Travel is measured from the visiting team's home stadium. None of these factors change the prediction apart from home field.
Where each team ranks among the 32 on its 2026 statistics before this game. A strength is top eight in the league; a weakness is bottom eight.
Ranks rest on 5 games a team, so they can move quickly. They describe what has happened so far, not the model's prediction, which also weighs last season and injuries.
Chance each player scores a rushing or receiving touchdown, and the price that would be fair for it.
If a bookmaker offers more than the fair price, the price is better than the model thinks it should be; most prices will be shorter because of the bookmaker's margin. Tested on 3,072 player-games from 2025: players rated 20–30% scored 26% of the time, those rated 40–50% scored 47%. Players ruled out are excluded and questionable players are flagged.
The bar covers the middle 80% of simulated games; the coloured mark is the midpoint. Wide bars are normal: yardage swings a lot from week to week.
Ranges come from testing these projections on 38,722 player-games in 2024 and 2025. The typical miss was 60 yards passing, 25 rushing and 23 receiving. Set your own over and under lines in the Parlay builder.
Combinations inside this game, priced from 60,000 simulated versions of it so that legs which move together are counted properly.
| Combination | Model chance | Fair price |
|---|---|---|
| Most likely pair49ers to win + Under 52.5 points | 40% 40% if independent | $2.52 |
| Most likely pair49ers to win + Over 52.5 points | 39% 39% if independent | $2.55 |
| Most likely trio49ers to win + Over 52.5 points + George Kittle anytime TD | 18% 15% if independent | $5.57 |
| Most likely trio49ers to win + Under 52.5 points + George Kittle anytime TD | 14% 15% if independent | $7.02 |
| Legs that move together49ers −7.5 + Over 52.5 points + George Kittle anytime TD | 12% 9.4% if independent | $8.37 |
| Legs that move togetherOver 52.5 points + Jayden Daniels anytime TD | 16% 13% if independent | $6.12 |
A win makes the over and big touchdown days more likely, so those legs reinforce each other. A favourite plus the under pull against each other. Bookmakers price same-game multis with their own margin, so what you are offered will be shorter than the fair price.
What each offence is expected to produce against the other defence. The stronger side in each row is in full colour.
Each figure is the league average plus that offence's own edge plus the opposing defence's edge, after pulling small samples toward the league average.
2026 season before this game, per game or per play.
Points per game each player is worth above a replacement, from pass rush, coverage, tackles for loss and production relative to teammates.
Active players only. Offensive linemen have no public statistics, so they are not rated here.
Points of margin lost this week from players on the injury report or injured reserve.
Out and Doubtful count as 0% available; Questionable counts as 50% to 75% depending on Friday practice. Players with no game status are treated as playing.
2026 results before this game, most recent last.
Regular-season games since 2022, most recent first.
The rest of the schedule played out 20,000 times from today's team ratings. The 49ers reach the playoffs most often, in 96% of those seasons.
The playoff field the 20,000 simulated seasons point to most often, and how often each team goes all the way. Every simulated season plays out the playoffs too: 2 v 7, 3 v 6, 4 v 5, then the top seed meets the lowest seed left.
The Jaguars would have the bye.
The 49ers would have the bye.
Seeds use records only; the league's tie-breakers are not modelled. Playoff games use each simulated season's ratings and home field for the higher seed; the Super Bowl is neutral.
Projected wins and the share of simulated seasons in which each team makes the playoffs or wins its division. Tap a team for its remaining games.
Seven teams from each conference qualify: four division winners and three wild cards. Ties on wins are split at random because the league's tie-breakers are not modelled. Ratings are allowed to be wrong by about 4 points a game in either direction, an assumption that has not been back-tested. Standings include Week 6; ratings are refreshed with those games when the next week is built.
The biggest moves in the model's power rank since last week, after Week 5's results and this week's injuries.
The average rating of each team's opponents, in points a game. Bars to the right mean tougher opponents.
Opponent ratings are today's, without short-term injuries, and do not adjust for home or away.
Pick any two teams and a venue to see who the ratings favour. It uses season ratings only, without this week's injury report.
Jaguars by 1.9, with 1.8 points of that from playing at home.
69 players were ruled out this week and 84 went in as questionable. The Ravens took the biggest hit from this week's report, 6.5 points of margin in the model.
The absences that moved the model most, the biggest names carrying a questionable tag, and the best news of the week. Written from the official reports, not from news articles.
Misses the game against the Browns (ankle). That costs the Ravens 6.5 points in the model.
Misses the game against the Panthers (calf). That costs the Eagles 5.7 points in the model.
Misses the game against the Steelers (thumb). That costs the Buccaneers 2.4 points in the model.
Misses the game against the Chargers (hamstring). That costs the Chiefs 1.3 points in the model.
Listed (illness) after he was limited on Friday. The model gave him a 60% chance of playing, which took 0.8 points off the Saints.
Listed (groin) after he did not practise on Friday. The model gave him a 45% chance of playing, which took 0.5 points off the Rams.
Was limited on Thursday (toe), was limited on Friday and carries no game status. The model rates him worth 1.8 points to the Ravens.
Did not practise on Thursday (hamstring), did not practise on Friday and carries no game status. The model rates him worth 1.7 points to the Commanders.
Was limited on Thursday (hamstring), was limited on Friday and carries no game status. The model rates him worth 1.5 points to the Commanders.
Players whose practice work changed between Thursday and the final report, ordered by how much the model values them.
None this week.
None this week.
Questionable players in finished games who then appeared in the box score, against the chance of playing the model assumed.
The marker is the model's assumed chance, measured on 2023–25 snap counts. A player counts as having played here only if he recorded a statistic, so backups who took a few snaps are missed and the real rates are somewhat higher. Offensive linemen, kickers and punters are left out because they do not record statistics. One week is too few players to judge the assumption.
Every club ordered by the points its injuries cost this week. The solid bar is this week's report and the lighter bar is players on injured reserve. Tap a team for its list.
No one with a game status.
Points are the model's cost at kick-off: the drop to the replacement multiplied by the chance the player misses. Players with no game status are treated as playing. The Cardinals and Falcons and Ravens and Bills and Panthers and Bears and Browns and Cowboys and Packers and Texans and Colts and Jaguars and Chiefs and Rams and Chargers and Raiders and Patriots and Saints and Giants and Jets and Eagles and Steelers and 49ers and Buccaneers and Titans and Commanders play on Monday night, so their statuses came out a day later than the model's inputs, and their costs were set from practice participation.
Every player rated 1 to 99: his place among the league's regular players at his position, so a 90 is as rare at corner as at receiver. Under it, the model's own figure: points a game he is worth over a replacement-level player.
Top 10 of the 32 ranked.
Top 10 of the 64 ranked.
Top 10 of the 132 ranked.
Top 10 of the 64 ranked.
Top 10 of the 103 ranked.
Top 10 of the 103 ranked.
Top 10 of the 103 ranked.
Top 10 of the 128 ranked.
Top 10 of the 108 ranked.
One page for every club: where it ranks, what it does well and badly, who its best players are, who is hurt and what is left on its schedule.
Record, the model's power rank with its move since last week, and the chance of making the playoffs. On a bye this week: Bengals and Lions and Dolphins and Vikings.
Every team on the measures that matter most. Ranks are 1st (best) to 32nd; efficiency is adjusted for opponents. Tap a heading to sort, or a team to open its page.
| Team | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| JAXJaguars | 1st | 6th | 10th | 24.6 | 14.0 | 7th | 8th | 7th | 3rd | +1.0 |
| LARams | 2nd | 13th | 9th | 23.8 | 24.0 | 14th | 12th | 17th | 13th | −0.6 |
| BUFBills | 3rd | 1st | 31st | 32.8 | 28.2 | 2nd | 5th | 29th | 23rd | −0.8 |
| NEPatriots | 4th | 11th | 7th | 22.0 | 18.8 | 9th | 23rd | 5th | 30th | −1.0 |
| SEASeahawks | 5th | 28th | 2nd | 26.8 | 21.0 | 17th | 30th | 2nd | 2nd | +0.0 |
| MINVikings | 6th | 30th | 1st | 21.8 | 12.8 | 24th | 24th | 1st | 1st | +1.0 |
| DETLions | 7th | 5th | 30th | 32.8 | 29.6 | 3rd | 9th | 28th | 25th | +0.8 |
| SF49ers | 8th | 2nd | 13th | 30.8 | 18.6 | 1st | 13th | 11th | 4th | +0.6 |
| HOUTexans | 9th | 26th | 6th | 22.0 | 22.4 | 15th | 26th | 19th | 9th | +0.6 |
| DENBroncos | 10th | 15th | 3rd | 19.4 | 22.2 | 23rd | 15th | 3rd | 28th | +0.4 |
| CINBengals | 11th | 7th | 15th | 25.6 | 21.8 | 11th | 14th | 6th | 24th | +0.6 |
| KCChiefs | 12th | 4th | 19th | 29.5 | 19.3 | 8th | 2nd | 14th | 10th | +1.0 |
| CHIBears | 13th | 8th | 23rd | 23.8 | 19.2 | 16th | 11th | 13th | 27th | −1.0 |
| GBPackers | 14th | 9th | 20th | 20.8 | 22.4 | 12th | 32nd | 10th | 26th | +0.6 |
| ATLFalcons | 15th | 18th | 8th | 23.8 | 24.4 | 27th | 3rd | 8th | 12th | −1.0 |
| PITSteelers | 16th | 16th | 17th | 22.2 | 21.8 | 26th | 4th | 15th | 20th | +0.2 |
| INDColts | 17th | 19th | 16th | 24.8 | 27.6 | 30th | 1st | 22nd | 32nd | −1.2 |
| DALCowboys | 18th | 3rd | 32nd | 27.6 | 27.2 | 4th | 10th | 32nd | 21st | −0.4 |
| CARPanthers | 19th | 12th | 14th | 30.3 | 27.3 | 6th | 22nd | 20th | 17th | +0.8 |
| LVRaiders | 20th | 14th | 21st | 26.4 | 25.8 | 10th | 29th | 21st | 6th | +1.0 |
| TBBuccaneers | 21st | 24th | 5th | 20.0 | 22.4 | 29th | 7th | 12th | 5th | −0.6 |
| CLEBrowns | 22nd | 20th | 12th | 18.6 | 22.4 | 18th | 31st | 4th | 29th | +0.0 |
| BALRavens | 23rd | 10th | 18th | 29.2 | 23.8 | 5th | 6th | 16th | 18th | +0.6 |
| LACChargers | 24th | 32nd | 4th | 16.8 | 25.8 | 32nd | 28th | 9th | 14th | −0.8 |
| NOSaints | 25th | 21st | 25th | 23.6 | 30.2 | 22nd | 16th | 26th | 31st | −1.2 |
| NYGGiants | 26th | 27th | 24th | 21.8 | 21.2 | 13th | 25th | 23rd | 22nd | +0.0 |
| NYJJets | 27th | 25th | 11th | 18.6 | 19.2 | 20th | 20th | 18th | 8th | +0.4 |
| MIADolphins | 28th | 23rd | 26th | 14.0 | 26.4 | 28th | 19th | 30th | 15th | −0.8 |
| WASCommanders | 29th | 22nd | 28th | 23.0 | 29.8 | 19th | 21st | 25th | 7th | +1.0 |
| ARICardinals | 30th | 17th | 29th | 21.6 | 32.4 | 21st | 18th | 31st | 11th | +0.2 |
| TENTitans | 31st | 31st | 27th | 11.6 | 21.8 | 31st | 17th | 27th | 16th | −0.6 |
| PHIEagles | 32nd | 29th | 22nd | 18.4 | 22.4 | 25th | 27th | 24th | 19th | −0.4 |
Points a game against an average team on a neutral field.
30th of 32 this week. Last week the Cardinals ranked 25th.
League rank on 2026 statistics through 5 games. Top eight is a strength, bottom eight a weakness.
Every measure with its league rank (1st is best) through 5 games. Efficiency is expected points added a play, adjusted for the opponents faced; green is top eight, red bottom eight.
The Cardinals' best by player rating (1 to 99, comparable across positions), with league rank at the position.
Every Cardinals player the model ranks: the starters and about one regular backup at each position, with league rank at the position. Offensive linemen, kickers and punters are not rated.
Costing the Cardinals 1.5 points this week.
12 games left. The schedule so far was the 7th hardest in the league; what is left is the 7th hardest.
Win chances use season ratings without this week's short-term injuries. Top seed in the conference: <1%.
Points a game against an average team on a neutral field.
15th of 32 this week. Last week the Falcons ranked 12th.
League rank on 2026 statistics through 5 games. Top eight is a strength, bottom eight a weakness.
Every measure with its league rank (1st is best) through 5 games. Efficiency is expected points added a play, adjusted for the opponents faced; green is top eight, red bottom eight.
The Falcons' best by player rating (1 to 99, comparable across positions), with league rank at the position.
Every Falcons player the model ranks: the starters and about one regular backup at each position, with league rank at the position. Offensive linemen, kickers and punters are not rated.
Costing the Falcons 0.4 points this week.
12 games left. The schedule so far was the 18th hardest in the league; what is left is the 12th hardest.
Win chances use season ratings without this week's short-term injuries. Top seed in the conference: 1%.
Points a game against an average team on a neutral field.
23rd of 32 this week. Last week the Ravens ranked 28th.
League rank on 2026 statistics through 5 games. Top eight is a strength, bottom eight a weakness.
Every measure with its league rank (1st is best) through 5 games. Efficiency is expected points added a play, adjusted for the opponents faced; green is top eight, red bottom eight.
The Ravens' best by player rating (1 to 99, comparable across positions), with league rank at the position.
Every Ravens player the model ranks: the starters and about one regular backup at each position, with league rank at the position. Offensive linemen, kickers and punters are not rated.
Costing the Ravens 6.5 points this week.
12 games left. The schedule so far was the 31st hardest in the league; what is left is the 14th hardest.
Win chances use season ratings without this week's short-term injuries. Top seed in the conference: 15%.
Points a game against an average team on a neutral field.
3rd of 32 this week. Last week the Bills ranked 6th.
League rank on 2026 statistics through 5 games. Top eight is a strength, bottom eight a weakness.
Every measure with its league rank (1st is best) through 5 games. Efficiency is expected points added a play, adjusted for the opponents faced; green is top eight, red bottom eight.
The Bills' best by player rating (1 to 99, comparable across positions), with league rank at the position.
Every Bills player the model ranks: the starters and about one regular backup at each position, with league rank at the position. Offensive linemen, kickers and punters are not rated.
Costing the Bills 0.0 points this week.
12 games left. The schedule so far was the 2nd hardest in the league; what is left is the 22nd hardest.
Win chances use season ratings without this week's short-term injuries. Top seed in the conference: 19%.
Points a game against an average team on a neutral field.
19th of 32 this week. Last week the Panthers ranked 16th.
League rank on 2026 statistics through 5 games. Top eight is a strength, bottom eight a weakness.
Every measure with its league rank (1st is best) through 5 games. Efficiency is expected points added a play, adjusted for the opponents faced; green is top eight, red bottom eight.
The Panthers' best by player rating (1 to 99, comparable across positions), with league rank at the position.
Every Panthers player the model ranks: the starters and about one regular backup at each position, with league rank at the position. Offensive linemen, kickers and punters are not rated.
Costing the Panthers 1.4 points this week.
And 1 more.
13 games left. The schedule so far was the 12th hardest in the league; what is left is the 8th hardest.
Win chances use season ratings without this week's short-term injuries. Top seed in the conference: 1%.
Points a game against an average team on a neutral field.
13th of 32 this week. Last week the Bears ranked 5th.
League rank on 2026 statistics through 5 games. Top eight is a strength, bottom eight a weakness.
Every measure with its league rank (1st is best) through 5 games. Efficiency is expected points added a play, adjusted for the opponents faced; green is top eight, red bottom eight.
The Bears' best by player rating (1 to 99, comparable across positions), with league rank at the position.
Every Bears player the model ranks: the starters and about one regular backup at each position, with league rank at the position. Offensive linemen, kickers and punters are not rated.
Costing the Bears 0.3 points this week.
And 1 more.
12 games left. The schedule so far was the 25th hardest in the league; what is left is the 1st hardest.
Win chances use season ratings without this week's short-term injuries. Top seed in the conference: 2%.
Points a game against an average team on a neutral field.
11th of 32 this week. Last week the Bengals ranked 14th.
League rank on 2026 statistics through 5 games. Top eight is a strength, bottom eight a weakness.
Every measure with its league rank (1st is best) through 5 games. Efficiency is expected points added a play, adjusted for the opponents faced; green is top eight, red bottom eight.
The Bengals' best by player rating (1 to 99, comparable across positions), with league rank at the position.
Every Bengals player the model ranks: the starters and about one regular backup at each position, with league rank at the position. Offensive linemen, kickers and punters are not rated.
Costing the Bengals 0.2 points this week.
12 games left. The schedule so far was the 14th hardest in the league; what is left is the 29th hardest.
Win chances use season ratings without this week's short-term injuries. Top seed in the conference: 5%.
Points a game against an average team on a neutral field.
22nd of 32 this week. Last week the Browns ranked 23rd.
League rank on 2026 statistics through 5 games. Top eight is a strength, bottom eight a weakness.
Every measure with its league rank (1st is best) through 5 games. Efficiency is expected points added a play, adjusted for the opponents faced; green is top eight, red bottom eight.
The Browns' best by player rating (1 to 99, comparable across positions), with league rank at the position.
Every Browns player the model ranks: the starters and about one regular backup at each position, with league rank at the position. Offensive linemen, kickers and punters are not rated.
Costing the Browns 0.1 points this week.
And 1 more.
12 games left. The schedule so far was the 20th hardest in the league; what is left is the 23rd hardest.
Win chances use season ratings without this week's short-term injuries. Top seed in the conference: 1%.
Points a game against an average team on a neutral field.
18th of 32 this week. Last week the Cowboys ranked 17th.
League rank on 2026 statistics through 5 games. Top eight is a strength, bottom eight a weakness.
Every measure with its league rank (1st is best) through 5 games. Efficiency is expected points added a play, adjusted for the opponents faced; green is top eight, red bottom eight.
The Cowboys' best by player rating (1 to 99, comparable across positions), with league rank at the position.
Every Cowboys player the model ranks: the starters and about one regular backup at each position, with league rank at the position. Offensive linemen, kickers and punters are not rated.
Costing the Cowboys 0.6 points this week.
12 games left. The schedule so far was the 22nd hardest in the league; what is left is the 21st hardest.
Win chances use season ratings without this week's short-term injuries. Top seed in the conference: 1%.
Points a game against an average team on a neutral field.
10th of 32 this week. Last week the Broncos ranked 9th.
League rank on 2026 statistics through 5 games. Top eight is a strength, bottom eight a weakness.
Every measure with its league rank (1st is best) through 5 games. Efficiency is expected points added a play, adjusted for the opponents faced; green is top eight, red bottom eight.
The Broncos' best by player rating (1 to 99, comparable across positions), with league rank at the position.
Every Broncos player the model ranks: the starters and about one regular backup at each position, with league rank at the position. Offensive linemen, kickers and punters are not rated.
Costing the Broncos 0.8 points this week.
11 games left. The schedule so far was the 1st hardest in the league; what is left is the 31st hardest.
Win chances use season ratings without this week's short-term injuries. Top seed in the conference: 7%.
Points a game against an average team on a neutral field.
7th of 32 this week. Last week the Lions ranked 10th.
League rank on 2026 statistics through 5 games. Top eight is a strength, bottom eight a weakness.
Every measure with its league rank (1st is best) through 5 games. Efficiency is expected points added a play, adjusted for the opponents faced; green is top eight, red bottom eight.
The Lions' best by player rating (1 to 99, comparable across positions), with league rank at the position.
Every Lions player the model ranks: the starters and about one regular backup at each position, with league rank at the position. Offensive linemen, kickers and punters are not rated.
Costing the Lions 0.0 points this week.
12 games left. The schedule so far was the 30th hardest in the league; what is left is the 19th hardest.
Win chances use season ratings without this week's short-term injuries. Top seed in the conference: 6%.
Points a game against an average team on a neutral field.
14th of 32 this week. Last week the Packers ranked 20th.
League rank on 2026 statistics through 5 games. Top eight is a strength, bottom eight a weakness.
Every measure with its league rank (1st is best) through 5 games. Efficiency is expected points added a play, adjusted for the opponents faced; green is top eight, red bottom eight.
The Packers' best by player rating (1 to 99, comparable across positions), with league rank at the position.
Every Packers player the model ranks: the starters and about one regular backup at each position, with league rank at the position. Offensive linemen, kickers and punters are not rated.
Costing the Packers 0.8 points this week.
And 6 more.
12 games left. The schedule so far was the 19th hardest in the league; what is left is the 2nd hardest.
Win chances use season ratings without this week's short-term injuries. Top seed in the conference: 2%.
Points a game against an average team on a neutral field.
9th of 32 this week. Last week the Texans ranked 13th.
League rank on 2026 statistics through 5 games. Top eight is a strength, bottom eight a weakness.
Every measure with its league rank (1st is best) through 5 games. Efficiency is expected points added a play, adjusted for the opponents faced; green is top eight, red bottom eight.
The Texans' best by player rating (1 to 99, comparable across positions), with league rank at the position.
Every Texans player the model ranks: the starters and about one regular backup at each position, with league rank at the position. Offensive linemen, kickers and punters are not rated.
Costing the Texans 0.0 points this week.
12 games left. The schedule so far was the 21st hardest in the league; what is left is the 27th hardest.
Win chances use season ratings without this week's short-term injuries. Top seed in the conference: <1%.
Points a game against an average team on a neutral field.
17th of 32 this week. Last week the Colts ranked 15th.
League rank on 2026 statistics through 5 games. Top eight is a strength, bottom eight a weakness.
Every measure with its league rank (1st is best) through 5 games. Efficiency is expected points added a play, adjusted for the opponents faced; green is top eight, red bottom eight.
The Colts' best by player rating (1 to 99, comparable across positions), with league rank at the position.
Every Colts player the model ranks: the starters and about one regular backup at each position, with league rank at the position. Offensive linemen, kickers and punters are not rated.
Costing the Colts 0.5 points this week.
12 games left. The schedule so far was the 8th hardest in the league; what is left is the 30th hardest.
Win chances use season ratings without this week's short-term injuries. Top seed in the conference: 1%.
Points a game against an average team on a neutral field.
1st of 32 this week. Last week the Jaguars ranked 1st.
League rank on 2026 statistics through 5 games. Top eight is a strength, bottom eight a weakness.
Every measure with its league rank (1st is best) through 5 games. Efficiency is expected points added a play, adjusted for the opponents faced; green is top eight, red bottom eight.
The Jaguars' best by player rating (1 to 99, comparable across positions), with league rank at the position.
Every Jaguars player the model ranks: the starters and about one regular backup at each position, with league rank at the position. Offensive linemen, kickers and punters are not rated.
Costing the Jaguars 0.1 points this week.
12 games left. The schedule so far was the 11th hardest in the league; what is left is the 32nd hardest.
Win chances use season ratings without this week's short-term injuries. Top seed in the conference: 27%.
Points a game against an average team on a neutral field.
12th of 32 this week. Last week the Chiefs ranked 7th.
League rank on 2026 statistics through 5 games. Top eight is a strength, bottom eight a weakness.
Every measure with its league rank (1st is best) through 5 games. Efficiency is expected points added a play, adjusted for the opponents faced; green is top eight, red bottom eight.
The Chiefs' best by player rating (1 to 99, comparable across positions), with league rank at the position.
Every Chiefs player the model ranks: the starters and about one regular backup at each position, with league rank at the position. Offensive linemen, kickers and punters are not rated.
Costing the Chiefs 1.9 points this week.
13 games left. The schedule so far was the 29th hardest in the league; what is left is the 6th hardest.
Win chances use season ratings without this week's short-term injuries. Top seed in the conference: 15%.
Points a game against an average team on a neutral field.
2nd of 32 this week. Last week the Rams ranked 4th.
League rank on 2026 statistics through 5 games. Top eight is a strength, bottom eight a weakness.
Every measure with its league rank (1st is best) through 5 games. Efficiency is expected points added a play, adjusted for the opponents faced; green is top eight, red bottom eight.
The Rams' best by player rating (1 to 99, comparable across positions), with league rank at the position.
Every Rams player the model ranks: the starters and about one regular backup at each position, with league rank at the position. Offensive linemen, kickers and punters are not rated.
Costing the Rams 1.0 points this week.
12 games left. The schedule so far was the 6th hardest in the league; what is left is the 20th hardest.
Win chances use season ratings without this week's short-term injuries. Top seed in the conference: 5%.
Points a game against an average team on a neutral field.
24th of 32 this week. Last week the Chargers ranked 22nd.
League rank on 2026 statistics through 5 games. Top eight is a strength, bottom eight a weakness.
Every measure with its league rank (1st is best) through 5 games. Efficiency is expected points added a play, adjusted for the opponents faced; green is top eight, red bottom eight.
The Chargers' best by player rating (1 to 99, comparable across positions), with league rank at the position.
Every Chargers player the model ranks: the starters and about one regular backup at each position, with league rank at the position. Offensive linemen, kickers and punters are not rated.
Costing the Chargers 1.6 points this week.
And 4 more.
12 games left. The schedule so far was the 9th hardest in the league; what is left is the 3rd hardest.
Win chances use season ratings without this week's short-term injuries. Top seed in the conference: <1%.
Points a game against an average team on a neutral field.
20th of 32 this week. Last week the Raiders ranked 19th.
League rank on 2026 statistics through 5 games. Top eight is a strength, bottom eight a weakness.
Every measure with its league rank (1st is best) through 5 games. Efficiency is expected points added a play, adjusted for the opponents faced; green is top eight, red bottom eight.
The Raiders' best by player rating (1 to 99, comparable across positions), with league rank at the position.
Every Raiders player the model ranks: the starters and about one regular backup at each position, with league rank at the position. Offensive linemen, kickers and punters are not rated.
Costing the Raiders 0.0 points this week.
12 games left. The schedule so far was the 24th hardest in the league; what is left is the 16th hardest.
Win chances use season ratings without this week's short-term injuries. Top seed in the conference: 1%.
Points a game against an average team on a neutral field.
28th of 32 this week. Last week the Dolphins ranked 29th.
League rank on 2026 statistics through 5 games. Top eight is a strength, bottom eight a weakness.
Every measure with its league rank (1st is best) through 5 games. Efficiency is expected points added a play, adjusted for the opponents faced; green is top eight, red bottom eight.
The Dolphins' best by player rating (1 to 99, comparable across positions), with league rank at the position.
Every Dolphins player the model ranks: the starters and about one regular backup at each position, with league rank at the position. Offensive linemen, kickers and punters are not rated.
Costing the Dolphins 0.3 points this week.
12 games left. The schedule so far was the 4th hardest in the league; what is left is the 5th hardest.
Win chances use season ratings without this week's short-term injuries. Top seed in the conference: <1%.
Points a game against an average team on a neutral field.
6th of 32 this week. Last week the Vikings ranked 11th.
League rank on 2026 statistics through 5 games. Top eight is a strength, bottom eight a weakness.
Every measure with its league rank (1st is best) through 5 games. Efficiency is expected points added a play, adjusted for the opponents faced; green is top eight, red bottom eight.
The Vikings' best by player rating (1 to 99, comparable across positions), with league rank at the position.
Every Vikings player the model ranks: the starters and about one regular backup at each position, with league rank at the position. Offensive linemen, kickers and punters are not rated.
A clean report this week.
12 games left. The schedule so far was the 28th hardest in the league; what is left is the 4th hardest.
Win chances use season ratings without this week's short-term injuries. Top seed in the conference: 25%.
Points a game against an average team on a neutral field.
4th of 32 this week. Last week the Patriots ranked 8th.
League rank on 2026 statistics through 5 games. Top eight is a strength, bottom eight a weakness.
Every measure with its league rank (1st is best) through 5 games. Efficiency is expected points added a play, adjusted for the opponents faced; green is top eight, red bottom eight.
The Patriots' best by player rating (1 to 99, comparable across positions), with league rank at the position.
Every Patriots player the model ranks: the starters and about one regular backup at each position, with league rank at the position. Offensive linemen, kickers and punters are not rated.
Costing the Patriots 1.3 points this week.
And 2 more.
12 games left. The schedule so far was the 3rd hardest in the league; what is left is the 25th hardest.
Win chances use season ratings without this week's short-term injuries. Top seed in the conference: 9%.
Points a game against an average team on a neutral field.
25th of 32 this week. Last week the Saints ranked 21st.
League rank on 2026 statistics through 5 games. Top eight is a strength, bottom eight a weakness.
Every measure with its league rank (1st is best) through 5 games. Efficiency is expected points added a play, adjusted for the opponents faced; green is top eight, red bottom eight.
The Saints' best by player rating (1 to 99, comparable across positions), with league rank at the position.
Every Saints player the model ranks: the starters and about one regular backup at each position, with league rank at the position. Offensive linemen, kickers and punters are not rated.
Costing the Saints 1.5 points this week.
12 games left. The schedule so far was the 5th hardest in the league; what is left is the 28th hardest.
Win chances use season ratings without this week's short-term injuries. Top seed in the conference: <1%.
Points a game against an average team on a neutral field.
26th of 32 this week. Last week the Giants ranked 27th.
League rank on 2026 statistics through 5 games. Top eight is a strength, bottom eight a weakness.
Every measure with its league rank (1st is best) through 5 games. Efficiency is expected points added a play, adjusted for the opponents faced; green is top eight, red bottom eight.
The Giants' best by player rating (1 to 99, comparable across positions), with league rank at the position.
Every Giants player the model ranks: the starters and about one regular backup at each position, with league rank at the position. Offensive linemen, kickers and punters are not rated.
Costing the Giants 2.1 points this week.
And 1 more.
12 games left. The schedule so far was the 32nd hardest in the league; what is left is the 17th hardest.
Win chances use season ratings without this week's short-term injuries. Top seed in the conference: 1%.
Points a game against an average team on a neutral field.
27th of 32 this week. Last week the Jets ranked 30th.
League rank on 2026 statistics through 5 games. Top eight is a strength, bottom eight a weakness.
Every measure with its league rank (1st is best) through 5 games. Efficiency is expected points added a play, adjusted for the opponents faced; green is top eight, red bottom eight.
The Jets' best by player rating (1 to 99, comparable across positions), with league rank at the position.
Every Jets player the model ranks: the starters and about one regular backup at each position, with league rank at the position. Offensive linemen, kickers and punters are not rated.
Costing the Jets 0.2 points this week.
12 games left. The schedule so far was the 27th hardest in the league; what is left is the 11th hardest.
Win chances use season ratings without this week's short-term injuries. Top seed in the conference: <1%.
Points a game against an average team on a neutral field.
32nd of 32 this week. Last week the Eagles ranked 24th.
League rank on 2026 statistics through 5 games. Top eight is a strength, bottom eight a weakness.
Every measure with its league rank (1st is best) through 5 games. Efficiency is expected points added a play, adjusted for the opponents faced; green is top eight, red bottom eight.
The Eagles' best by player rating (1 to 99, comparable across positions), with league rank at the position.
Every Eagles player the model ranks: the starters and about one regular backup at each position, with league rank at the position. Offensive linemen, kickers and punters are not rated.
Costing the Eagles 6.4 points this week.
And 2 more.
12 games left. The schedule so far was the 16th hardest in the league; what is left is the 26th hardest.
Win chances use season ratings without this week's short-term injuries. Top seed in the conference: <1%.
Points a game against an average team on a neutral field.
16th of 32 this week. Last week the Steelers ranked 18th.
League rank on 2026 statistics through 5 games. Top eight is a strength, bottom eight a weakness.
Every measure with its league rank (1st is best) through 5 games. Efficiency is expected points added a play, adjusted for the opponents faced; green is top eight, red bottom eight.
The Steelers' best by player rating (1 to 99, comparable across positions), with league rank at the position.
Every Steelers player the model ranks: the starters and about one regular backup at each position, with league rank at the position. Offensive linemen, kickers and punters are not rated.
Costing the Steelers 0.8 points this week.
12 games left. The schedule so far was the 10th hardest in the league; what is left is the 18th hardest.
Win chances use season ratings without this week's short-term injuries. Top seed in the conference: 2%.
Points a game against an average team on a neutral field.
5th of 32 this week. Last week the Seahawks ranked 2nd.
League rank on 2026 statistics through 5 games. Top eight is a strength, bottom eight a weakness.
Every measure with its league rank (1st is best) through 5 games. Efficiency is expected points added a play, adjusted for the opponents faced; green is top eight, red bottom eight.
The Seahawks' best by player rating (1 to 99, comparable across positions), with league rank at the position.
Every Seahawks player the model ranks: the starters and about one regular backup at each position, with league rank at the position. Offensive linemen, kickers and punters are not rated.
Costing the Seahawks 0.4 points this week.
11 games left. The schedule so far was the 17th hardest in the league; what is left is the 9th hardest.
Win chances use season ratings without this week's short-term injuries. Top seed in the conference: 4%.
Points a game against an average team on a neutral field.
8th of 32 this week. Last week the 49ers ranked 3rd.
League rank on 2026 statistics through 5 games. Top eight is a strength, bottom eight a weakness.
Every measure with its league rank (1st is best) through 5 games. Efficiency is expected points added a play, adjusted for the opponents faced; green is top eight, red bottom eight.
The 49ers' best by player rating (1 to 99, comparable across positions), with league rank at the position.
Every 49ers player the model ranks: the starters and about one regular backup at each position, with league rank at the position. Offensive linemen, kickers and punters are not rated.
Costing the 49ers 0.2 points this week.
12 games left. The schedule so far was the 13th hardest in the league; what is left is the 24th hardest.
Win chances use season ratings without this week's short-term injuries. Top seed in the conference: 51%.
Points a game against an average team on a neutral field.
21st of 32 this week. Last week the Buccaneers ranked 26th.
League rank on 2026 statistics through 5 games. Top eight is a strength, bottom eight a weakness.
Every measure with its league rank (1st is best) through 5 games. Efficiency is expected points added a play, adjusted for the opponents faced; green is top eight, red bottom eight.
The Buccaneers' best by player rating (1 to 99, comparable across positions), with league rank at the position.
Every Buccaneers player the model ranks: the starters and about one regular backup at each position, with league rank at the position. Offensive linemen, kickers and punters are not rated.
Costing the Buccaneers 2.6 points this week.
12 games left. The schedule so far was the 15th hardest in the league; what is left is the 15th hardest.
Win chances use season ratings without this week's short-term injuries. Top seed in the conference: <1%.
Points a game against an average team on a neutral field.
31st of 32 this week. Last week the Titans ranked 31st.
League rank on 2026 statistics through 5 games. Top eight is a strength, bottom eight a weakness.
Every measure with its league rank (1st is best) through 5 games. Efficiency is expected points added a play, adjusted for the opponents faced; green is top eight, red bottom eight.
The Titans' best by player rating (1 to 99, comparable across positions), with league rank at the position.
Every Titans player the model ranks: the starters and about one regular backup at each position, with league rank at the position. Offensive linemen, kickers and punters are not rated.
Costing the Titans 0.0 points this week.
12 games left. The schedule so far was the 26th hardest in the league; what is left is the 10th hardest.
Win chances use season ratings without this week's short-term injuries. Top seed in the conference: <1%.
Points a game against an average team on a neutral field.
29th of 32 this week. Last week the Commanders ranked 32nd.
League rank on 2026 statistics through 5 games. Top eight is a strength, bottom eight a weakness.
Every measure with its league rank (1st is best) through 5 games. Efficiency is expected points added a play, adjusted for the opponents faced; green is top eight, red bottom eight.
The Commanders' best by player rating (1 to 99, comparable across positions), with league rank at the position.
Every Commanders player the model ranks: the starters and about one regular backup at each position, with league rank at the position. Offensive linemen, kickers and punters are not rated.
A clean report this week.
12 games left. The schedule so far was the 23rd hardest in the league; what is left is the 13th hardest.
Win chances use season ratings without this week's short-term injuries. Top seed in the conference: <1%.
17 of 31 winners so far, 1 behind the market. 31 games is far too few to prove an edge either way: a coin-flipper gets a record like this often. This page exists so the record is public, whatever it turns out to be.
Picks grouped by how confident the model was. The bar is how often those picks won; the marker is how often the model said they would.
A well-calibrated model has each bar ending at its marker. With this few games each group can be off by 20 points or more through luck alone.
The model's record each week against the market's. Open a past week to see every prediction as it stood, with the results.
| Week | Winners | Market | Against the spread | Totals | Brier, model v market | |
|---|---|---|---|---|---|---|
| Week 4 | 9 of 15 | 9 of 15 | 11 of 13 | 8 of 15 | 0.183 v 0.212 | Open Week 4 → |
| Week 5 | 8 of 15 | 9 of 15 | 6 of 14 | 7 of 15 | 0.245 v 0.239 | Open Week 5 → |
| Week 6 | 0 of 1 | 0 of 1 | 1 of 1 | 0 of 1 | 0.271 v 0.340 | this week |
The picks that aged best and worst so far.
They won by 5. The model missed the margin by 0.1 points.
Week 4 · JAX 22, CIN 17 ✓The market had the Bengals as favourites. The model gave the Jaguars 67% and they won.
Week 4 · JAX 22, CIN 17 ✓The model's most confident pick that lost. The Ravens won by 7.
Week 5 · BAL 30, ATL 23 ✕The model had the Bears by 4.6. The game finished CHI 7, GB 31.
Week 5 · CHI 7, GB 31 ✕The model's pick and confidence, the final score, and whether the pick, its side of the spread and its side of the total were right.
| Game | Model pick | Chance | Final | Pick | Spread | Total |
|---|---|---|---|---|---|---|
| SEA @ DEN | Seahawks by 0.7 | 52% | 14–17 | ✕ | ✓ | ✕ |
| Game | Model pick | Chance | Final | Pick | Spread | Total |
|---|---|---|---|---|---|---|
| TB @ DAL | Cowboys by 5.8 | 67% | 24–16 | ✕ | ✓ | ✕ |
| PHI @ JAX | Jaguars by 10.1 | 77% | 17–19 | ✓ | ✕ | ✓ |
| CHI @ GB | Bears by 4.6 | 63% | 7–31 | ✕ | ✕ | ✓ |
| CIN @ MIA | Bengals by 4.1 | 62% | 31–24 | ✓ | push | ✓ |
| LV @ NE | Patriots by 4.8 | 64% | 17–45 | ✓ | ✓ | ✕ |
| MIN @ NO | Vikings by 2.6 | 58% | 23–13 | ✓ | ✓ | ✕ |
| CLE @ NYJ | Browns by 0.5 | 51% | 12–17 | ✕ | ✕ | ✕ |
| IND @ PIT | Steelers by 1.1 | 53% | 22–34 | ✓ | ✕ | ✕ |
| HOU @ TEN | Texans by 5.3 | 65% | 26–3 | ✓ | ✕ | ✕ |
| NYG @ WAS | Giants by 0.0 | 50% | 27–27 | ✕ | ✓ | ✓ |
| DEN @ LAC | Broncos by 3.0 | 59% | 23–17 | ✓ | ✕ | ✕ |
| DET @ ARI | Lions by 4.2 | 62% | 45–21 | ✓ | ✕ | ✓ |
| SF @ SEA | Seahawks by 2.4 | 57% | 32–29 | ✕ | ✓ | ✓ |
| BAL @ ATL | Falcons by 7.9 | 72% | 30–23 | ✕ | ✕ | ✓ |
| BUF @ LA | Rams by 2.2 | 56% | 37–34 | ✕ | ✓ | ✕ |
| Game | Model pick | Chance | Final | Pick | Spread | Total |
|---|---|---|---|---|---|---|
| IND @ WAS | Colts by 5.8 | 70% | 30–13 | ✓ | ✓ | ✕ |
| TEN @ BAL | Ravens by 8.1 | 76% | 18–24 | ✓ | ✓ | ✓ |
| NE @ BUF | Bills by 5.4 | 69% | 29–26 | ✕ | ✓ | ✕ |
| NYJ @ CHI | Bears by 10.3 | 82% | 12–23 | ✓ | ✓ | ✕ |
| JAX @ CIN | Jaguars by 4.9 | 67% | 22–17 | ✓ | ✓ | ✓ |
| DAL @ HOU | Texans by 3.6 | 63% | 34–30 | ✕ | ✕ | ✓ |
| ARI @ NYG | Cardinals by 1.7 | 56% | 24–36 | ✕ | ✓ | ✓ |
| LA @ PHI | Rams by 7.0 | 73% | 24–20 | ✓ | ✓ | ✕ |
| GB @ TB | Buccaneers by 1.0 | 54% | 17–14 | ✕ | push | ✓ |
| MIA @ MIN | Vikings by 11.9 | 85% | 10–15 | ✓ | ✕ | ✓ |
| KC @ LV | Chiefs by 3.6 | 63% | 30–27 | ✓ | ✓ | ✕ |
| LAC @ SEA | Seahawks by 11.0 | 83% | 23–30 | ✓ | push | ✕ |
| DEN @ SF | 49ers by 4.7 | 66% | 14–24 | ✓ | ✓ | ✓ |
| DET @ CAR | Lions by 1.1 | 54% | 26–32 | ✕ | ✓ | ✓ |
| ATL @ NO | Saints by 1.7 | 56% | 45–24 | ✕ | ✓ | ✕ |
Predictions are fixed before kick-off from the final injury reports. "Spread" and "Total" compare the model's side with the market line at the time.
Every week from 2012 to 2025 replayed through the model, using only what was known at the time. Settings were tuned on 2012 to 2020 and judged on 2021 to 2025, seasons the tuning never saw.
| Held-out seasons 2021 to 2025, 1,359 games | Winners picked | Brier score | Margin miss | Model's side v the spread | Total-points miss |
|---|---|---|---|---|---|
| Model before the back-test | 62.8% | 0.2266 | 10.21 | 49.5% | 11.66 |
| Model after the changes | 64.7% | 0.2184 | 10.05 | 49.6% | 10.47 |
| Market closing line | 66.7% | 0.2117 | 9.75 | – | 10.31 |
| Average miss, lower is better | Before | After | Player-games |
|---|---|---|---|
| Touchdown scorers (Brier score) | 0.1598 | 0.1572 | 17,283 |
| Rushing yards | 25.02 | 24.42 | 3,761 |
| Receiving yards | 22.41 | 22.19 | 10,859 |
| Passing yards | 60.52 | 59.28 | 2,072 |
These are simulated results, not live picks. Every week was predicted using only games already played. The back-test covers the team model and the quarterback term; the full injury model could not be replayed, though a separate check supported its size. Lower is better for the Brier score and both miss columns.
Each week the assumptions are checked against the results. A setting only changes when several seasons of evidence move it, not on one week alone.