Two engines, no takes. The Decision Index grades every 4th-down and PAT call by win probability at the moment of decision — outcome-blind, branch math shown. The cap boards price every skill contract against the production it actually bought. Every claim carries the number that would prove it wrong.
How often the league went for it on competitive 4th downs last season — against how often the math said go. The gap is free wins. Mike McDaniel left the fewest behind (0.12 expected wins lost all season).
Fourth downs and PATs, valued three ways by win probability at the snap — go, kick, punt — and graded on the choice, never the result. A failed go that was plus-3% is a good call. The league went for it on 19.8% of competitive fourth downs last season; the math wanted 36.9%. That gap is free wins, and some coaches leave more of them on the table than others.
| COACH | WINS LOST | GO% | MODEL GO% | AGREE |
|---|---|---|---|---|
| Mike McDaniel · MIA | 0.12 | 14% | 30% | 82% |
| Dan Campbell · DET | 0.12 | 26% | 36% | 85% |
| Nick Sirianni · PHI | 0.13 | 12% | 20% | 90% |
| Sean McDermott · BUF | 0.18 | 30% | 35% | 75% |
| Brian Schottenheimer · DAL | 0.19 | 29% | 42% | 76% |
| Zac Taylor · CIN | 0.20 | 14% | 34% | 73% |
| Sean McVay · LA | 0.22 | 28% | 38% | 82% |
| Andy Reid · KC | 0.22 | 27% | 41% | 82% |
| COACH | WINS LOST | GO% | MODEL GO% | AGREE |
|---|---|---|---|---|
| Brian Daboll · NYG | 0.56 | 26% | 45% | 74% |
| DeMeco Ryans · HOU | 0.54 | 13% | 37% | 71% |
| Kevin Stefanski · CLE | 0.45 | 15% | 37% | 73% |
| Ben Johnson · CHI | 0.44 | 22% | 36% | 72% |
| Dan Quinn · WAS | 0.42 | 23% | 43% | 71% |
| Mike Tomlin · PIT | 0.42 | 17% | 38% | 76% |
| Brian Callahan · TEN | 0.40 | 11% | 35% | 67% |
| John Harbaugh · BAL | 0.37 | 17% | 38% | 75% |
Every grade shows the win probability of each option. The reader sees the arithmetic, not just the verdict.
Surplus = production percentile minus pay percentile, inside each position. A restructure is a balance-transfer credit card; dead money is rent on an apartment you moved out of; a rookie deal is a below-market lease with a known expiration. The boards below are why teams hunt the third one.
| PLAYER | CAP | PROD | PAY | +/− |
|---|---|---|---|---|
| Devon Achane · RB · Dolphins | 1% | 93 | 49 | +44 |
| Kyle Pitts · TE · Falcons | 1.5% | 86 | 47 | +40 |
| Travis Kelce · TE · Chiefs | 1.6% | 95 | 56 | +40 |
| Mac Jones · QB · 49ers | 0.9% | 57 | 24 | +33 |
| Derrick Henry · RB · Ravens | 1.8% | 92 | 69 | +23 |
| A.J. Brown · WR · Patriots | 2% | 88 | 65 | +23 |
| Trey McBride · TE · Cardinals | 2.7% | 100 | 79 | +21 |
| James Conner · RB · Cardinals | 1.5% | 82 | 62 | +20 |
| PLAYER | CAP | PROD | PAY | +/− |
|---|---|---|---|---|
| Harold Fannin Jr. · TE · Browns · R | 0.5% | 88 | 16 | +72 |
| Ladd McConkey · WR · Chargers · R | 0.9% | 83 | 35 | +49 |
| Brock Bowers · TE · Raiders · R | 1.6% | 98 | 56 | +42 |
| Drake Maye · QB · Patriots · R | 2.8% | 96 | 55 | +41 |
| Zay Flowers · WR · Ravens · R | 1.4% | 94 | 55 | +39 |
| Puka Nacua · WR · Rams · R | 1.9% | 97 | 60 | +38 |
| Bo Nix · QB · Broncos · R | 1.7% | 80 | 43 | +37 |
| Tyler Warren · TE · Colts · R | 1.6% | 93 | 56 | +37 |
| PLAYER | CAP | PROD | PAY | +/− |
|---|---|---|---|---|
| Deshaun Watson · QB · Browns | 12.1% | 8 | 90 | -82 |
| Brandon Aiyuk · WR · 49ers | 4% | 4 | 85 | -82 |
| Evan Engram · TE · Broncos | 4.7% | 23 | 98 | -74 |
| T.J. Hockenson · TE · Vikings | 5% | 28 | 100 | -72 |
| Chris Godwin · WR · Buccaneers | 10.6% | 29 | 100 | -71 |
| Cole Kmet · TE · Bears | 2.5% | 16 | 77 | -60 |
| Darius Slayton · WR · Giants | 5.3% | 31 | 90 | -59 |
| Mike Gesicki · TE · Bengals | 2.4% | 21 | 74 | -53 |
Production is what happened on the field, injury seasons included — the board measures realized value, not talent. Skill positions only: public data can't grade a guard's contract, so we don't pretend it can. R = rookie-scale deal.
Every restructure moves today's bill to later years. This is each team's balance-transfer balance — the money already committed against caps that haven't arrived yet.
Outgoing versus incoming coach, on expected wins lost per season to 4th-down and PAT decisions (league average: 0.38). Promoted coordinators have no head-coach record — they're priced at league average, labeled, and graded from their first real Sunday. Scheme, development, and leadership are not in this number, and we say so.
| TEAM | OUT → IN | EWL OUT | EWL IN | Δ WINS |
|---|---|---|---|---|
| PIT | Mike Tomlin → Mike McCarthy | 0.49 | 0.3 | +0.19W |
| NYG | Brian Daboll → John Harbaugh | 0.49 | 0.31 | +0.17W |
| LV | Pete Carroll → Klint KubliakLEAGUE-AVG PRIOR | 0.4 | 0.38 | +0.01W |
| TEN | Brian Callahan → Robert Saleh | 0.45 | 0.48 | -0.03W |
| CLE | Kevin Stefanski → Todd MonkenLEAGUE-AVG PRIOR | 0.35 | 0.38 | -0.04W |
| MIA | Mike McDaniel → Jeff HafleyLEAGUE-AVG PRIOR | 0.32 | 0.38 | -0.06W |
| BAL | John Harbaugh → Jesse MinterLEAGUE-AVG PRIOR | 0.31 | 0.38 | -0.07W |
Two honest numbers per move. Market wins: what the contract says the team believes (pay above position replacement, through the league's own exchange rate, calibrated on the QB market — the one position where public data measures point contribution directly). Production wins: what the last two seasons actually support. The gap between them is the bet the team is making. These are priced estimates, not measurements — and we say so.
| MOVE | MARKET | PROD |
|---|---|---|
| Trent McDuffie · CB · Chiefs → Rams | +2.59W | — |
| Laremy Tunsil · LT · Texans → Commanders | +2.51W | — |
| Jaelan Phillips · ED · Eagles → Panthers | +2.5W | — |
| Trey Hendrickson · ED · Bengals → Ravens | +2.33W | — |
| Tyler Linderbaum · C · Ravens → Raiders | +2.24W | — |
| Odafe Oweh · ED · Chargers → Commanders | +1.98W | — |
| Malik Willis · QB · Titans → Dolphins | +1.85W | — |
| Tytus Howard · RT · Texans → Browns | +1.85W | — |
| TEAM | MARKET NET | COMP '27 | HEADLINE |
|---|---|---|---|
| Raiders | +7.33W | — | added Tyler Linderbaum (C) |
| Commanders | +5.9W | — | added Laremy Tunsil (LT) |
| Rams | +3.96W | — | added Trent McDuffie (CB) |
| Titans | +3.96W | — | added John Franklin-Myers (ED) |
| Browns | +3.22W | — | added Tytus Howard (RT) |
| Eagles | -2.78W | +0.47W | lost Jaelan Phillips (ED) |
| Packers | -2.83W | +0.59W | lost Romeo Doubs (WR) |
| Colts | -3.29W | +0.75W | lost Michael Pittman, Jr. (WR) |
| Seahawks | -3.4W | +0.83W | lost Boye Mafe (ED) |
| Chiefs | -4.46W | +1.22W | lost Trent McDuffie (CB) |
Where spending has actually lined up with winning (2013-2025): pass-rush cap share correlates most (r +0.151); offensive-line share, essentially not at all (r -0.01). Correlation, not destiny — but it prices the conventional wisdom.
COMP '27 is the formula paying teams back: projected 2027 compensatory picks for net free-agent losses, priced on the market-graded chart. Best haul: Chiefs (R3 R4 R5 R6 ≈ +1.22W, headlined by Trent McDuffie (CB, $31M)). Rounds are ±1 — the league's exact cutoffs aren't public, and we say so.
Instead of borrowing the 1991 trade card the league still uses, we graded every slot by the second contract its players eventually commanded — the market's own verdict, busts counted at zero. The gap at the top is the trade-up tax: the old card prices pick 1 at five late-firsts; a decade of outcomes says two and a half.
| PICK | 2ND-DEAL CAP% | WINS | ARCLINE | 1991 CARD |
|---|---|---|---|---|
| #1 | 10.06% | 2.63W | 1000 | 100 |
| #16 | 5.37% | 1.41W | 534 | 33 |
| #32 | 4.09% | 1.07W | 407 | 20 |
| #48 | 3.19% | 0.84W | 317 | 14 |
| #64 | 2.96% | 0.77W | 294 | 9 |
| #100 | 1.79% | 0.47W | 178 | 3 |
| #150 | 1.15% | 0.30W | 114 | 1 |
| #200 | 0.68% | 0.18W | 67 | 0 |
| #250 | 0.51% | 0.13W | 51 | 0 |
The 1991 card scores it 1032 for 1700— a fleecing, don't do it. Market-graded outcomes score the same trade at +0.12W — essentially fair. A decade of second contracts says two good day-two picks really do equal one top-five swing.
Pick 1 vs pick 32: market-graded 2.46× · 1991 card 5.08×. Teams still trading with the old card systematically overpay to move up — the gap is the tax, and on draft night we grade every trade against both charts, live.
Every big veteran deal gets a full underwriting: the price against the production, the aging curve's expectation by the final year, and the escape hatch — the first season the team can leave for cap-positive savings, and what it costs. Verdicts lock at publish. Each one carries the concrete result that would prove it wrong and a review date — and when the season delivers evidence, the grade lands here, hit or miss. Nobody else self-grades takes. That's the product.
$64M a year is the 98th percentile of the QB market for 92th-percentile production — the price and the output live on the same block. Fair deals don't make headlines; they make payrolls work.
Wrong in either direction: top-5 production makes it a bargain; falling out of the top 30 makes it an overpay.
$55M a year is the 96th percentile of the QB market for 88th-percentile production — the price and the output live on the same block. Fair deals don't make headlines; they make payrolls work.
Wrong in either direction: top-5 production makes it a bargain; falling out of the top 30 makes it an overpay.
$50M a year puts Will Anderson Jr. at the 100th percentile of the ED market, with 67% of the deal guaranteed. Public data can't grade a ED's film the way it grades a receiver's production — so we price the deal, say so out loud, and watch the player.
Graded on price discipline only: this entry is wrong if the ED market's top of book moves so far past this deal by 2027 that it reads like a bargain in hindsight.
Daniel Jones arrives at the 94th percentile of the QB market — $44M a year, 56% guaranteed — against 69th-percentile production over the last two seasons. The gap between the bill and the output has to be closed by a career year. That's the bet, and it's priced like a certainty.
Wrong if Daniel Jones finishes top-5 in QB production in 2026 — a top-5 year makes this price fair and this verdict a miss.
$42.15M a year is the 100th percentile of the WR market for 99th-percentile production — the price and the output live on the same block. Fair deals don't make headlines; they make payrolls work.
Wrong in either direction: top-5 production makes it a bargain; falling out of the top 30 makes it an overpay.
$40.1M a year puts Danielle Hunter at the 99th percentile of the ED market, with 99% of the deal guaranteed. Public data can't grade a ED's film the way it grades a receiver's production — so we price the deal, say so out loud, and watch the player.
Graded on price discipline only: this entry is wrong if the ED market's top of book moves so far past this deal by 2027 that it reads like a bargain in hindsight.
The GM layer runs year-round. The weekly engine — player projections, DFS optimizer, survivor strategy, live decision calls — arrives with September. Every model walks forward before it publishes; the picks gate stays closed until the numbers earn it.