Here is the September 2026 board in one sentence: NFL teams are simultaneously leaving wins on the field through conservative fourth-down decisions and leaving cap room on the field through contracts that price hope rather than production. Both problems are measurable, both are persistent, and neither is a secret anymore — which makes them more interesting, not less.

Across 32,737 graded decisions in the 2025 season, the patterns hold. The coaches who trust the math add a fraction of a win per game — modest by the play, enormous by October. The contracts that ignore the gap between what a player earns and what a player produces create obligations that follow a front office for years. The bill always arrives; it just arrives wearing a different uniform than the one you were imagining.

The ledger this month carries 100 verdicts: 21 fair prices, 11 overpays, and 68 watches — positions where public data can grade the deal but not the player. Zero good values, which is worth saying plainly. The market has gotten efficient enough at the top that true bargains now live almost exclusively on rookie contracts. More on that below.

In 2025, NFL teams went for it on 4th down in competitive situations 18.9% of the time. The model, built on play-by-play data since 1999, says the right number is 35.6%. That gap — 16.7 percentage points of decisions left on the table — translates into expected wins lost (EWL): the cumulative win probability surrendered across a season by choosing the cautious option when the math favors aggression. Think of it as the difference between your retirement account's projected balance and its actual balance, calculated choice by choice.

Dan Campbell and Detroit sit at 0.103 EWL on the season — roughly a tenth of a win lost to suboptimal decisions across 139 graded situations, good for 0.61 points of win probability added per game. He went for it 26.3% of the time against a model recommendation of 35.4%, and agreed with the model's call on 85.9% of decisions. That agreement rate is the signal. Campbell isn't the most aggressive coach in the league; he's closer to the most calibrated one. Nick Sirianni in Philadelphia posted the lowest EWL among the three leaders at 0.130, adding 0.76 win-probability points per game across 144 decisions — the best per-game number of the group — while going for it only 12.1% of the time against a model recommendation of 18.7%. When your recommendation is that modest and you still hit 89.7% agreement, the decisions you're getting right are the ones that matter most.

On the other end, Brian Daboll in New York lost 0.543 expected wins — more than half a win — with a 72.8% model-agreement rate and a go-rate of 24.3% against a recommended 40.8%. DeMeco Ryans in Houston was worse on agreement (67.2%) while going for it only 9.8% of the time in situations where the model wanted 36.1%. That is not a rounding error; that is a philosophical disagreement with the arithmetic. Kevin Stefanski in Cleveland lost 0.425 expected wins across 151 decisions, the most of any coach on the list, at 71% agreement. The week's sharpest single call came from Arizona: Week 6, fourth-and-seven at the opponent's nine-yard line, down four, Jonathan Gannon chose to go for it — win probability at 20.5% — when a field goal was worth 33.3%. The cost was 12.8 percentage points of win probability, the steepest single-decision loss of the month. A field goal ties the game. Going for it and converting wins it outright, which is the appeal — but the math charges a steep premium for that appeal, and the right answer there is the kick.

Patrick Mahomes: $64 million a year, the 98th percentile of the quarterback market, against 92nd-percentile production. Verdict: fair price. The thesis is simple — the price and the output live on the same block, and fair deals don't make headlines; they make payrolls work. The falsifier is equally simple: top-five production makes it a bargain; falling out of the top 30 makes it an overpay. We re-grade in March 2027. At 31 years old with an aging score of 0.52 — well below the threshold that would flag a decline concern — this one is about as comfortable as a long-term quarterback deal gets. The exit cost in 2029 is $31.3 million in dead money, which is the number that matters if the comfortable deal becomes uncomfortable.

Daniel Jones in Indianapolis is the instructive contrast. $44 million a year, the 94th-percentile of the quarterback market, 56% guaranteed — against 69th-percentile production over the last two seasons. Verdict: overpay. The gap between the bill and the output has to be closed by a career year, and the contract prices that career year like a certainty. The falsifier here is honest: if Jones finishes top-five in quarterback production in 2026, this verdict is wrong, and we'll say so plainly in March. His aging score of 1.06 is the number worth watching — it sits above the threshold that historically signals production risk rather than growth. The exit in 2027 carries $35.2 million in dead money, which means the Colts are essentially renting this contract whether it works or not.

Jaxon Smith-Njigba at $42.15 million a year is the 100th percentile of the wide receiver market against 99th-percentile production. Fair price — and a useful illustration of how the ledger works. Top-of-market deals look alarming in isolation; they look reasonable when the production percentile is one tick below the pay percentile. The falsifier is the same as Mahomes: top-five output makes it a bargain, falling out of the top 30 makes it an overpay, re-grade in March. What the verdicts share — fair, overpay, or watch — is a date on the calendar and a clear condition for being wrong. A verdict without a falsifier is just an opinion wearing a number's clothes.

A rookie contract is a below-market lease with a known expiration date. The team locks in a rate set before the player proved anything; the player performs at a level the open market would pay two or three times more to replicate. The surplus — the gap between what a player earns and what his production would cost on the open market — is the real currency of roster construction. The teams that stack it tend to win in January.

Harold Fannin Jr. in Cleveland is the month's clearest example. He's carrying $1.53 million against the cap — the 16th percentile of tight end pay — while producing at the 88th percentile of tight end production. Surplus of 72 percentile points. The Browns are paying 16 cents on the dollar for 88-cent production, which is the definition of a windfall. Drake Maye in New England sits at $9.99 million, the 54th percentile of quarterback pay, against 96th-percentile production — a 42-point surplus. Brock Bowers in Las Vegas adds 98th-percentile tight end production on $4.95 million, the 56th percentile of tight end pay, for a 42-point surplus of his own. The rookie window is not a loophole; it is the mechanism the league built into the system, and the teams that identify and develop those players are the ones who can afford to pay veterans fairly when the window closes.

Among veterans, Devon Achane in Miami is the standout. $3.15 million — 1% of the cap, 49th-percentile running back pay — against 93rd-percentile production. A 44-point surplus on a position the market routinely undervalues. Travis Kelce in Kansas City runs at $4.90 million (56th-percentile pay) against 95th-percentile production, a 40-point surplus at age 37. Kyle Pitts in Atlanta has quietly become a similar story: 86th-percentile production at $4.62 million, 47th-percentile pay, 40-point surplus. The Falcons drafted him to be that player and eventually paid the price when he wasn't; now that he is, they're getting the production at a discount. That's how it goes sometimes.

Future proration is the balance-transfer balance on a credit card: the restructured money didn't disappear, it just moved to a later statement. The Texans lead the league with $891.5 million in future proration commitments while carrying their quarterback at only 6.8% of the cap currently — meaning the bill for past restructures is large and the quarterback line is temporarily suppressed. The Cowboys sit at $868 million in future obligations with Dak Prescott at 14.2% of the cap, one of the highest quarterback cap percentages in the league. The 49ers carry $734.3 million with their quarterback line at 8.5%. The Saints show $704.8 million with a quarterback line of just 1.9% — which either means they've been disciplined at the position or they haven't solved it yet, and those are very different problems wearing the same number.

The albatross list is where the bill has already arrived. Deshaun Watson in Cleveland: $40.96 million against the cap, 90th-percentile pay, 8th-percentile production. A surplus of negative 83 points — the worst on the board. The Browns are simultaneously paying Watson like an elite quarterback and watching Harold Fannin Jr. produce like one on $1.53 million. Brandon Aiyuk in San Francisco carries the same negative-83 surplus: $13.67 million at 87th-percentile wide receiver pay against 4th-percentile production. Chris Godwin in Tampa Bay costs $33.68 million — 10.6% of the cap, the 100th percentile of wide receiver pay — against 28th-percentile production, a negative-72 surplus. Evan Engram in Denver: $14.14 million, 98th-percentile tight end pay, 23rd-percentile production, negative-74 surplus.

The common thread across the albatross list isn't bad luck, though injury is part of several of these stories — and where realized value diverges from talent, we try to say so. The common thread is that the contracts were priced on a projection rather than a track record, and when the projection didn't land, the team still owed the money. Dead money is rent on an apartment you moved out of: the obligation runs whether or not anyone is sleeping there. The Texans, Cowboys, and 49ers are all managing large future proration balances while trying to compete now. How they handle the math over the next two offseasons will tell you more about their rosters in 2028 than anything that happens on the field this fall.