Buffalo Still Has the Engine — Now Check the Tires

The first thing to know about Buffalo's 2025 season is that it was almost exactly what it deserved to be. The Bills went 12-5, and their point differential — plus-116 on the year — translates to 11.7 structural wins using our fitted points-per-win model. The luck delta is a rounding error: +0.3, meaning they won about a third of a game more than their underlying performance forecast. This was not a mirage season and it was not a stolen one. It was a 12-win team that won 12 games.

The one-score record adds a little texture. Buffalo went 5-3 in games decided by one score, which is a winning mark but not an outlier — and it barely moves the needle when the structural number is already sitting at 11.7. Close-game records are among the noisiest statistics in the sport; the point differential is the signal. When those two things agree this tightly, you're looking at a team that earned its position in the conference. The Bills didn't catch breaks to get here. They were simply better than almost everyone they played.

The offense was excellent and the passing game was the reason. Buffalo posted a +0.20 EPA per pass attempt in 2025 — that's the kind of number that shows up when a quarterback and his supporting cast are operating above the league's average expectation on nearly every dropback. The run game was modestly positive at +0.08 EPA per carry, a useful complement without being the thing that drives the machine. Josh Allen is the machine. The rushing efficiency says everything else is contextual.

The defense was a more complicated story. Pass defense held up well: opponents generated just -0.11 EPA per attempt against Buffalo's secondary, which means the Bills were actively taking value away from opposing passing games. The run defense was the sore spot — a +0.09 EPA per carry allowed, meaning opponents were generating positive value on the ground at a rate that should concern a team with AFC championship aspirations. That number is a known problem, and the offseason addressed it directly.

Situationally, the Bills were efficient where it matters most. A 52.3% conversion rate on third and fourth downs is a real edge — roughly the difference between sustaining drives and punting away possessions — and a red-zone EPA per play of +0.15 means scoring opportunities were being converted into actual points rather than field goals. Head coach Sean McDermott posted a decision index of 75.9% agreement with our model, going for it on fourth down at a 29.1% rate against a modeled suggestion of 31.6%. That 2.5-point gap cost Buffalo about 0.18 expected wins last season — a small but real inefficiency from a coach who is otherwise managing the game well.

The offseason had a clear logic to it. The biggest arrival is Bradley Chubb, the edge defender who comes over from Miami on a three-year deal at $14.5 million per year — a 91st-percentile price tag for the position that buys 1.16 market wins of expected production. He is 30 years old, and we've flagged this one WATCH, not for the deal itself but for the aging trajectory. The exit ramp is manageable: $6 million dead in 2028 if the Bills need to use it. At center, Connor McGovern arrives from the Jets at $13.1 million annually on a four-year contract — a 96th-percentile price for the position and another WATCH verdict. Buffalo is betting that it needed to spend at the top of the market on interior protection, and the logic is defensible; the price discipline is something to monitor as the market resets over the next two years. Corner Dee Alford rounds out the free-agent class at $5.25 million per year, adding competition in a secondary that the cap profile shows is significantly below market investment.

On the departure side, guard David Edwards left for New Orleans at $15.25 million annually — the most expensive player to walk out the door — and backup quarterback Mitchell Trubisky signed with Tennessee at $5.25 million. The net of all free-agent movement works out to +0.97 market wins on the production ledger, essentially a push on the production side but a meaningful gain in positional need fulfillment, particularly at edge rusher.

The draft class is anchored by two second-round picks. Clemson defensive end T.J. Parker at pick 35 carries a 1.07-win chart value — that's a pick in the range where Day 1 contributors are common, and he arrives on a roster that has a clear path to snaps at a position of need. Ohio State corner Davison Igbinosun at pick 62 adds 0.79 chart wins to a secondary the Bills have been visibly underinvesting in, per the cap profile. The rest of the class — Jude Bowry (OT), Skyler Bell (WR), Kaleb Elarms-Orr (LB), Jalon Kilgore (CB), Zane Durant (DT), Toriano Pride Jr. (CB), Tommy Doman (P), and Ar'maj Reed-Adams (G) — contributes chart value in the 0.13-to-0.45 range, the range where depth and development are the honest expectation.

On balance, this offseason helps — carefully, in places that needed it, with price risks that are real but not roster-threatening. The Chubb signing is the most consequential move, and the case for it is the run-defense EPA. Buffalo was giving up value on the ground consistently last season. Adding a veteran pass rusher with Chubb's track record changes the calculus for opposing coordinators in ways that don't always show up cleanly in EPA splits but matter in January. The age and the price are the honest concerns; they're flagged, not hidden.

The cap picture has two clear problem spots, and they're worth naming plainly. Tight end Dawson Knox is eating 2.8% of the cap at the 81st percentile of pay while producing at the 30th percentile of the position. Wide receiver Josh Palmer is at 3.9% of the cap — 85th-percentile pay — for 36th-percentile production. Together, those two contracts represent real opportunity cost in a league where the margin between good and great roster construction is often one or two high-value signings. The Bills aren't buried, but they're paying full price for below-average returns at two skill positions, and that limits the flexibility they'd otherwise have.

The bargain that holds everything together is James Cook. He's producing at the 90th percentile of running backs while taking up just 2.0% of the cap at the 74th percentile of pay. That kind of efficiency — production outrunning price by 16 percentile points — is genuinely uncommon, and it matters most when the team around him needs to spend elsewhere. The Bills have built wisely where it counts most: the franchise quarterback is at 14.7% of the cap, and everything downstream of that contract looks sustainable for now.

The cap profile's biggest z-score imbalance tells the larger story: quarterback investment is 1.51 standard deviations above league average, cornerback investment is 0.97 below, and safety is 0.87 below. The Igbinosun and Alford additions are an acknowledgment of that gap, but they don't close it in one offseason. Buffalo is a team that has built from the middle of the roster outward and is now trying to shore up the edges. The direction is right. The job isn't done.

The arithmetic here is clean, even if the result isn't dramatic. The 2025 point differential produces a structural base of 11.7 wins. The offseason moves net to zero on the production side — the arrivals replace departures without a clear upgrade on a wins-produced basis, though the market-wins logic is more favorable. The coaching staff is unchanged, so there's no decision-index delta to apply. The accounting projection for 2026 is 11.7 structural wins — the same read as last year's underlying performance. This is an accounting projection from the ledger, not a pick, not a bet, and not a forecast that claims to know what Josh Allen's ankle or Stefon Diggs's absence or a January snowstorm will mean in January.

The falsifier is straightforward: if the Bills finish 2026 with nine wins or fewer, this read was wrong in a way that matters. That result would suggest either that the roster's known vulnerabilities — the run defense, the albatross contracts, the aging edge signing — compounded in ways the structural model couldn't capture, or that regression from the close-game record arrived harder than expected. A result in the 10-to-12 range confirms the thesis. Thirteen or more wins would mean the offseason additions outperformed their market-wins projections, which is possible — it's just not the base case.

The vacated opportunity heading into 2026 is modest. Mitchell Trubisky, who departed for Tennessee, accumulated 31.1 PPR points across 9 carries in 2025 — a backup quarterback's footprint, not a target to get excited about. The carries redistribute to the active roster without a clear fantasy consequence worth pricing in.

The sleeper case worth building is James Cook, and it rests on the efficiency argument. He's operating at the 90th percentile of running back production while the market — reflected in his 74th-percentile pay — hasn't priced that in. Managers who let name recognition drive running back valuation instead of per-opportunity numbers tend to undervalue the players in efficient offenses who accumulate work without the hype. Cook qualifies. He's the right back on the right team at a price that hasn't caught up to what he's actually doing.

There are no aging-flag fades from this brief that apply to the fantasy-relevant skill positions. The flags attached to Chubb are a team-building concern, not a weekly DFS concern — edge rushers don't show up in your lineup. The albatross exposure at tight end and wide receiver (Knox and Palmer) is more relevant: both are being paid like contributors and producing like role players, which means neither commands the target share their cap number implies. Fade both relative to their draft-cost expectations and find efficiency elsewhere on the depth chart.

Everything above traces to a live instrument, and the receipts stay public: the GM boards carry Sean McDermott's decision card, the team's cap posture ($204M already charged to future caps), and every contract verdict with its falsifier and review date — including Bradley Chubb (WATCH), Connor McGovern (WATCH). When the season grades these reads, the grades post whether they flatter us or not.