SharpEdge

The Ledger

This is the accountability issue. Every play graded this week was bet at DraftKings at -110 for one unit. Units: a fixed bet size so results can be compared. One unit is one unit whether that is $10 or $100. At -110 a winner pays 0.91 units and a loser costs the full unit. That is why a split card still loses money.

The sides were spread bets. The spread is the head start the underdog gets. Seahawks -3 means Seattle has to win by more than 3 for that bet to cash. The Colts-Commanders play was a total: the over/under on combined points from both teams.

The rest of this issue runs on two terms. Closing line value, or CLV, is whether the line you bet was better than where the market closed. It is the best long-term sign a bettor knows what they are doing, because closing lines are the sharpest numbers there are. Hit rate is the percentage of bets that won. At standard -110 odds you need 52.4% just to break even.

Running totals:

That is a 28.6% hit rate against a 52.4% break-even, and it has cost 6.36 units. Fourteen graded plays is a small sample, and regression works in both directions. Regression means pulling an extreme number back toward average because small samples lie. A 6-0 team is good; it is not undefeated-good. A 4-10 card probably isn't this bad either. That is not an excuse. It is the reason the next section matters more than this one.

The three published plays without a grade show up in the pack as blank "none" entries from weeks 3 and 4. They have no game, no side and no line, so there is nothing to grade.

What the Closing Line Said

• Beat the close: Texans -2.5, closed -3.0. That is half a point better, and the half point sits on the good side of a key number. 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.
• Matched the close: Browns +2.5, closed +2.5.
• Lost to the close: Panthers +3.5, closed +4.5. Over 47.5, closed 46.5. We gave up a full point on each.

This list tells you more than the win column does. A handful of games tells you almost nothing, because one tipped pass flips the record. The closing number is the market's final price, set after the sharp money, the injury reports and every model in the building have had their say. If you keep buying better than that price, the wins follow. It works like my walk-in cooler. If I buy beef Monday for less than it costs Friday, I bought well, even if Friday's dinner rush gets rained out. The rush is variance. The price is the skill.

By that measure the week was worse than the record. Only the Texans bet beat the close, and both winners were bought at the closing number or worse. The season's +0.18 average CLV is positive, but barely. It says the prices have been roughly fair and the results have been bad. It does not prove the process works. It only says the process isn't obviously broken.

The Lean, Revisited

Texans -2.5, loss. We had the right number and got the wrong result. The market agreed with us and kept moving to -3.0, so I held the good side of 3. A Houston win by exactly 3 cashes my ticket but only pushes at the closing number. The pack doesn't include the final score, so I won't invent a story about how it got away. Verdict: variance. Same bet, same number, next time.

Over 47.5, Colts at Commanders, loss. This one is on me. I bought 47.5 and the market closed at 46.5, a full point the other way. That is a bad price whether the game goes over or not, and it didn't. Verdict: the number was wrong, and probably the read too. There is also a sizing problem. This was a Lean, and it risked the same full unit as a Play. So did the Browns. If a Lean is the weaker opinion, betting it like a strong one is a flaw in the ledger itself, win or lose.

Panthers +3.5, win. Mostly luck. The side held, but the market closed at +4.5, so I bought early and on the wrong side of the move.

Browns +2.5, win. Luck. I bought right at the closing number, so there was no edge to claim. A coin flip that landed.

Now the cold water from the backtest. It is walk-forward, 2023-2025, graded against closing spreads, and it counts only plays where the model disagreed with the market by 2+ points. Over 437 plays it went 51.7% against the spread, which is under break-even:

• Edges of 2-3 points hit 54.7%.
• Edges of 3-4.5 points hit 52.1%.
• Edges of 4.5 or more hit 47.7%.

The bigger the disagreement, the more likely the model is the one that's wrong. The edge you can repeat is 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. This pack only covers DraftKings and no other books were checked, so that free edge is sitting on the table. You should know that.

Looking Ahead

These are the games where the DraftKings opening line landed on a key number or a half point off one. The model column comes from power ratings. A power rating 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. This model counts home field as 1.8 points. Every injury list in the pack is empty for this slate, so nothing below depends on injury news.

Eagles at Jaguars (Sunday Oct 11, 9:30 AM ET) didn't open near a key number but moved onto one. It opened JAX -4.5 and now sits on 7, a 2.5-point move that went straight past the model's JAX -6.2. The value is gone.

Watch list (not plays, nothing logged):

• Giants +3 at Washington. The model leans Giants by 2.6 points. That is the only edge size that cleared break-even in the backtest, and it sits exactly on 3. What would make me wrong: the Giants score 18.2 points a game, and underdogs that can't score don't cover.
• Titans +7 vs Houston. The move from 6.5 to 7 gave Titans backers the key number. The model leans Tennessee by 2.1, and Houston is 0-4 against the spread this season. What would make me wrong: Tennessee scores 13.5 a game and has the league's lowest power rating at -5.5.

Passes:

• Bengals at Dolphins. The model loves Miami by 6.0 points. That puts it in the 4.5-plus bucket, the one that hit 47.7%, and Miami scores 11.2 points a game. I'm treating the model as wrong here until it proves otherwise.
• Broncos at Chargers. The model leans LAC by 3.4, which is the 52.1% bucket, under break-even. The Chargers score 16.0 a game.
• Bears at Packers. The move to 3 helped the Packers side, and the model leans Green Bay by 2.0. But Chicago scores 27.8 and allows 15.5, while Green Bay scores 17.5 and allows 26.2. I don't trust the lean.
• Ravens at Falcons. It opened BAL -3.5 and is now -2.5, straight through 3 toward Atlanta. The model still leans Atlanta by 2.2, but the +3.5 is gone, and whoever bet the opening line owns the CLV. Chasing a number after it crosses 3 is the Panthers lesson again.
• Colts at Steelers. The model leans Indianapolis by 1.7, which is under the 2-point floor the backtest uses.
• 49ers at Seahawks. A 0.4 lean is no lean. If you want Seattle anyway, the nflverse line feed shows -3.0, so DraftKings' -2.5 is the better side of 3.
• Bills at Rams. The model total is 47.9 and DraftKings has 54.5. The backtest covers spreads, not totals, and our only total this week lost a full point to the close. I'm staying away.
• Browns at Jets, Raiders at Patriots. The pack has no current DraftKings number for either game, so there is no price to judge. On the Patriots, the model says NE -7.5 while its own scoring projection has the Raiders at 23.4 and the Patriots at 18.6. When the model argues with itself, I side with the market.

Ledger Entries

Plain English

• Units: A fixed bet size so results can be compared. One unit is one unit whether that is $10 or $100.
• 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.
• 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.
• Hit rate: The percentage of bets that won. At standard -110 odds you need 52.4% just to break even.
• Regression: Pulling an extreme number back toward average because small samples lie. A 6-0 team is good; it is not undefeated-good.
• 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.
• 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.

Sources

• DraftKings lines, opening lines and closing numbers, via ESPN.
• nflverse play-by-play, schedule and line data.
• Official injury reports.
• The SharpEdge ledger and power rating model, including its 2023-2025 walk-forward backtest.

21+. Gambling problem? Call 1-800-GAMBLER.