SharpEdge

The Card

Sizing. Bet in units, which means a fixed bet size so results can be compared. One unit is one unit whether that is $10 or $100. We bet one unit on a Play and half a unit on a Lean, and nothing on a Pass. The ledger is 0-5 so far, so we are not pressing. We are not chasing either. The size stays flat.

Model vs Market

Every spread here is the point spread: the head start the underdog gets. Seahawks -3 means Seattle has to win by more than 3 for that bet to cash. The total is the over/under on combined points from both teams. The model line comes from a power rating, which is a number for how good a team is, in points, against an average team on a neutral field. Subtract two ratings, add home field, and you have a model spread.

First, the plain truth. Our backtest ran from 2023 to 2025 on closing spreads. When the model disagreed with the market by 2 or more points, it went 437 plays at 51.7% against the spread. Break-even is 52.4%. The model does not beat the market on its own. It tells us where to look. The small gaps of 2 to 3 points hit 54.7%. Gaps of 3 to 4.5 hit 52.1%. Gaps of 4.5 or more hit 47.7%. So the biggest disagreements on this board are the ones we trust least.

That kills KC @ MIA (8.4), CIN @ PIT on the side (7.2), and PHI @ CHI (5.6). Our ratings are built mostly on last season, blended 80/20 with the first two weeks and then pulled toward average by regression. Regression means pulling an extreme number back toward average because small samples lie. A 6-0 team is good; it is not undefeated-good. That cuts both ways. Miami's line has given up sacks on 15.3% of dropbacks, and the defense has a 0.0% sack rate. The market may be reacting to what it has seen, and our ratings may be too slow. Chicago is starting Tyson Bagent per the pack's projected quarterbacks, and a rating built on last season does not know who is under center.

Two more flags. BAL @ DAL: the pack lists the venue as Maracana Stadium, not Dallas's home field. The model adds 1.8 points of home field to every home team. If that edge does not apply, most of the 3.0-point gap goes away. Pass. NYJ @ DET: the model says Detroit by 8.7, but its own points projection has the Jets at 27.8 and the Lions at 21.5. When a model argues with itself, we sit out. Detroit -6.5 lays the points on the right side of 7, and the Lions have a 10-day rest edge. Rest edge means days since each team's last game. Teams off a bye or a Thursday game have a real, measurable rest advantage. Even so, the Jets' profile ranks 6th in net EPA and Detroit's defense ranks 28th. Pass.

Line Shop

The only book on the card today is DraftKings. The multi-book comparison table starts next issue.

Here is why it matters. Line shopping means checking the same bet at several sportsbooks and taking the best number. Getting +3.5 instead of +3 is free equity, and it costs nothing but a minute. It matters most around key numbers: NFL games land on 3 and 7 far more than any other margin, so a half point on either side of those numbers is worth real money.

Four DraftKings spreads sit right on a key number this week: LAC @ BUF (7), SEA @ WAS (7), HOU @ IND (3), and NE @ JAX (3). Our two spread bets are both at +3. On +3, a 3-point loss is a push and you get your money back. On +3.5, the same game is a win. If you find New England or Indianapolis at +3.5 anywhere, take it over the DraftKings number.

One thing to watch. On KC @ MIA, DraftKings has the total at 45.5, and the nflverse data feed carries 46.5. That feed is not a book you can bet into. Still, when two sources disagree by a full point, the number is not settled. Shop the total before you touch it.

The Lean

NE @ JAX, Sunday Sep 27, 1:00 PM ET, EverBank Stadium (outdoors, grass). Drake Maye at Trevor Lawrence. Our ratings have New England 2nd in the league at 4.8 and Jacksonville 5th at 3.6. That makes the Patriots the better team getting 3 points on the road. The gap is 2.3 points, which is the backtest's best bucket.

Start with EPA per play, or Expected Points Added: how many points a play moved the team toward scoring compared with an average play from the same spot. Positive is good on offense, negative is good on defense. New England's defense is at -0.182, 5th in the league, and it has allowed 6.5 points a game. Its explosive play rate allowed is 3.5%. Explosive play rate means the share of plays that gained 20 or more yards. Explosives are how bad teams win and good teams cover. Jacksonville's offense is real, though: 0.142 EPA, 8th, with a 51.9% success rate. Success rate is the share of plays that kept the offense on schedule (roughly 40% of the yards needed on 1st down, 60% on 2nd, all of it on 3rd and 4th). The Patriots' offense is the problem: -0.092 EPA, 25th, 2.5 turnovers a game, and sacks on 9.8% of dropbacks against a Jaguars defense that sacks on 8.6%.

Both teams went 12-5 against the spread last season. Jacksonville beat the number by 6.7 points on average, and New England by 6.2. The implied team total is how many points the betting market expects a team to score, worked out from the spread and the total. Bigger number, more touchdowns to go around. It has Jacksonville at 24.2 and New England at 21.2. DraftKings has New England +130 on the moneyline, which is a bet on who wins, no spread. -166 means risk 166 to win 100; +142 means risk 100 to win 142. The model total is 34.7 against 45.5. That 10.8-point gap is too big to bet, but it points the same way as the side: a low-scoring game where getting 3 points matters.

Call: NE +3, Play. What would make me wrong: Maye keeps giving the ball away at 2.5 a game, and Jacksonville's 51.9% success rate turns into short fields.

CIN @ PIT, Sunday Sep 27, 1:00 PM ET, Acrisure Stadium (outdoors, grass). A division game with Joe Burrow at Aaron Rodgers. Cincinnati's defense ranks 2nd at -0.222 EPA and Pittsburgh's ranks 3rd at -0.202. Pittsburgh's offense ranks 31st at -0.330, with a 35.2% success rate and 11.5 points a game. The Steelers still have a 68% pass rate, which is the share of plays the offense throws instead of runs. It tells you which teams feed receivers and which feed backs. Their passing EPA is -0.411. They throw a lot and it goes badly.

The Bengals' defense stops the run at -0.303 EPA per rush. Pittsburgh's defense sacks the quarterback on 13.7% of dropbacks, and Cincinnati's line allows 8.3%. Implied totals: Cincinnati 23.0, Pittsburgh 19.5. Model total: 38.3. Pittsburgh is 0-2 to the under this year.

A note on honesty: the backtest covers spreads only. This Play rests on the football, not on the model's record. On the side, Pittsburgh +3.5 gets through the 3, which is the right side of the key number. But a 7.2-point gap is exactly what the backtest tells us not to trust. We are leaving the side alone.

Call: Under 42.5, Play. What would make me wrong: Pittsburgh's defense has allowed an explosive play rate of 9.3%. Two long Cincinnati plays can wreck an under.

HOU @ IND, Sunday Sep 27, 1:00 PM ET, Lucas Oil Stadium (roof status not listed in the pack). C.J. Stroud at Daniel Jones, a division game, both teams on 7 days' rest. The model has this nearly even, Indy by 0.6, against Houston -3. Indianapolis runs well (0.121 EPA per rush) and protects Jones: a 3.1% sack rate allowed against a Houston defense that sacks at 10.3%. Houston's offense ranks 26th at -0.092 with a 39.6% success rate, even though it runs 74.5 plays a game.

That is the case for Indy. The case against is ugly. Indy's defense ranks 32nd at 0.269 EPA and allows a 52.9% success rate. Houston beat the spread by 4.5 points on average last year. The model's 7.6-point over on the total is exactly the kind of number we do not bet.

Call: IND +3, Lean. I like the side more than the number. +3.5 would make this a Play. What would make me wrong: Houston's volume against the league's worst defense turns into a two-score margin.

Quick hit, ATL @ GB, Thursday Sep 24, 8:15 PM ET, Lambeau Field. Atlanta's offense ranks 32nd at -0.403 EPA, scores 8.0 points a game, and turns it over 3.5 times a game. Green Bay's ranks 29th. Both teams are on 4 days' rest. Green Bay listed guards/tackles Aaron Banks and Zach Bako-Bewele (knee), WR Jayden Reed (neck), and DT Warren Brinson (calf) as did not participate. Atlanta listed DE Samson Ebukam (hamstring) and CB A.J. Terrell (groin) the same way. None has a game status yet. Under 43.5, Lean. What would make me wrong: Atlanta turnovers hand Green Bay short fields.

The Ledger

It has been a bad start, and it goes on the page as it is. The hit rate is the percentage of bets that won. At standard -110 odds you need 52.4% just to break even. We are at 0.0%. Closing line value is CLV: whether the line you bet was better than where the market closed. The best long-term sign a bettor knows what they are doing, because closing lines are the sharpest numbers there are. Ours averages 0.4, which means that on average we beat the close. Five games is noise. Beating the close is the signal that lasts. Same reminder as always: the model alone does not beat the market in the backtest. What we are selling is line shopping, key numbers, and knowing which side of a number the football supports. Every Play and Lean in this issue goes on the ledger.

Ledger Entries

Plain English

• Spread: The head start the underdog gets. Seahawks -3 means Seattle has to win by more than 3 for that bet to cash.
• Total: The over/under on combined points from both teams.
• Moneyline: A bet on who wins, no spread. -166 means risk 166 to win 100; +142 means risk 100 to win 142.
• Key numbers: NFL games land on 3 and 7 far more than any other margin, so a half point on either side of those numbers is worth real money.
• Line shopping: Checking the same bet at several sportsbooks and taking the best number. Getting +3.5 instead of +3 is free equity, and it costs nothing but a minute.
• Units: A fixed bet size so results can be compared. One unit is one unit whether that is $10 or $100.
• Power rating: A number for how good a team is, in points, against an average team on a neutral field. Subtract two ratings, add home field, and you have a model spread.
• Regression: Pulling an extreme number back toward average because small samples lie. A 6-0 team is good; it is not undefeated-good.
• Rest edge: Days since each team's last game. Teams off a bye or a Thursday game have a real, measurable rest advantage.
• EPA per play: Expected Points Added: how many points a play moved the team toward scoring compared with an average play from the same spot. Positive is good on offense, negative is good on defense.
• Success rate: The share of plays that kept the offense on schedule (roughly 40% of the yards needed on 1st down, 60% on 2nd, all of it on 3rd and 4th).
• Pass rate: The share of plays the offense throws instead of runs. It tells you which teams feed receivers and which feed backs.
• Explosive play rate: The share of plays that gained 20 or more yards. Explosives are how bad teams win and good teams cover.
• Implied team total: How many points the betting market expects a team to score, worked out from the spread and the total. Bigger number, more touchdowns to go around.
• Hit rate: The percentage of bets that won. At standard -110 odds you need 52.4% just to break even.
• Closing line value: CLV: whether the line you bet was better than where the market closed. The best long-term sign a bettor knows what they are doing, because closing lines are the sharpest numbers there are.

Sources

Lines: DraftKings via ESPN. Team profiles and the second line/total feed: nflverse play-by-play. Injuries: official injury reports. Ratings, model lines, and backtest: SharpEdge model.

21+. Analysis and opinion, not betting advice. If you or someone you know has a gambling problem, call 1-800-GAMBLER.