Dallas Spent Big and Barely Moved the Needle

The Cowboys finished 7-9-1 last season, which is a number that tells you almost nothing useful on its own. The number that matters is -40 — that's their point differential for the year, and when you run it through the fitted points-per-win formula, you get a structural win total of 7.4. The actual record landed right on top of that. The luck delta was +0.1. In other words, Dallas wasn't unlucky, wasn't lucky, and wasn't better than the scoreboard suggested. They were exactly as good as they looked.

The one-score record — 4-3 in games decided by eight points or fewer — adds a small asterisk. A team that goes 4-3 in close games and still finishes below .500 is telling you the margin games went their way more often than not. The underlying production simply wasn't there to support a better record even with that modest edge. When the point differential and the close-game results both tell the same story, the story is usually right: this was a 7-win football team.

That's the honest baseline heading into 2026. Not a contender hiding behind bad luck, not a pretender propped up by fortune. Just a team that needs to get better to win more games, which is the least interesting sentence in football analysis and also, here, the most accurate one.

The passing offense was real. At +0.18 EPA per attempt, Dallas ranked comfortably among the league's better passing attacks on a per-play basis. That's a legitimate strength — the kind of number that tells you the quarterback and the skill positions were doing something right when the ball was in the air. The rushing game, at -0.02 EPA per carry, was essentially neutral, which is fine as long as you're not asking it to carry the offense. Dallas wasn't.

The defense is where the story turns. Opponents produced +0.25 EPA per passing attempt against this group — that's a significant number, and not in a good way. The rush defense wasn't sharp either, giving up +0.08 EPA per carry. A team allowing positive EPA on both run and pass plays on a per-snap basis is going to have a hard time winning consistently, and Dallas didn't. The situational numbers reinforce it: a red zone EPA per play of -0.06 means that when opponents got close, they scored. Frequently enough to matter.

The conversion rate on third and fourth down — 47.3% — was a genuine positive, suggesting the offense could move the chains when it needed to. But converting third downs while the defense bleeds points is a treadmill, not a winning formula. On the decision-making side, Brian Schottenheimer agreed with the model on 76.1% of his fourth-down calls, going for it 29.3% of the time against a modeled rate of 42.4%. That gap cost Dallas an estimated 0.194 wins — not a catastrophic number, but it's a fifth of a win left on the field by being too conservative at the line of scrimmage.

The offseason additions are modest in quantity and significant in concentration. On the free-agent side, Dallas signed safety Jalen Thompson away from Arizona at $11 million per year — a deal the market prices at 0.85 wins of expected production. The production baseline from play-by-play isn't yet available for Thompson, so the win value here is market-implied rather than confirmed. No players of note departed. The net on production wins is zero; the market-implied net is +0.85, which is the most optimistic honest reading of this offseason.

The new contracts are the more interesting story. George Pickens signed a one-year deal at $27.3 million per year — 98th percentile pay, 95th percentile production last season, age 26. The verdict is FAIR PRICE, and that's the right call: you're paying for a receiver who earned it, on a short enough runway that both sides retain leverage. The falsifier is clean — top-five production makes it a bargain; falling outside the top 30 makes it an overpay. One year limits the exposure either way. Rashan Gary comes over on a two-year deal at $16 million annually, landing in the 92nd percentile of edge rusher pay at age 29. The production percentile isn't yet established in Dallas's scheme, which is why this one earns a WATCH rather than a verdict. There's an exit in 2028 with $8.7 million in dead money, which is a meaningful but not ruinous number if the pass rush doesn't materialize.

The draft class is the most defensible part of the offseason. Caleb Downs at pick 11 carries 1.5 chart wins — the largest single investment, and the position pairing with Thompson signals Dallas is serious about rebuilding the back end of the defense. Malachi Lawrence at pick 23 adds another 1.18 chart wins at defensive end. Jaishawn Barham in the third round (0.54 chart wins), Drew Shelton at offensive tackle (0.38), Devin Moore at corner (0.37), and LT Overton at edge (0.35) round out a class that's heavily weighted toward defensive investment. The full class totals 4.52 chart wins — a reasonable return on seven picks, with the top-end value concentrated where the team's biggest structural problem lives.

The honest answer is: marginally help, with a wide range of outcomes. The draft class added real positional investment in the secondary and along the defensive line — the exact places the EPA profile said the team was hemorrhaging production. Whether Caleb Downs and Malachi Lawrence contribute meaningfully as rookies is the single largest swing variable in the projection. First-round picks at those positions in that range historically take a year to find their footing, which means the benefit may arrive in 2027 more reliably than 2026.

The cap picture adds texture. Dallas is carrying $249.9 million in future proration — a significant deferred obligation — and the positional spending is skewed heavily toward receiver, where the group Z-score sits at +1.99. That's the most receiver-heavy concentration in the league on a relative basis. The quarterback group is also above market at +1.19. Meanwhile, edge is at -1.12 — below market, which is an interesting tension given the Gary signing and the draft investment there. Dak Prescott is consuming 13.1% of the cap, George Pickens 8.2%, Quinnen Williams 6.5%, and CeeDee Lamb 5.9%. That's four players at roughly a third of the cap, before you get to the rest of the roster.

The one quiet bright spot in the cap structure is Javonte Williams — 74th percentile production at 57th percentile pay, 1.2% of the cap. That's the kind of deal that doesn't make headlines and keeps an offense functional. There are no albatrosses on the books, which is a genuine asset when the rest of the spending is this concentrated. But the arithmetic here is less about whether the Cowboys hurt themselves and more about whether a heavily receiver-invested, pass-first team can fix a defense that gave up positive EPA on every play type last year. The additions suggest an attempt; the results are unconfirmed.

This is an accounting projection from the ledger — not a betting model, not a pick, and not a promise. Walk the arithmetic: the structural base from the 2025 point differential is 7.4 wins. The offseason adds zero confirmed production wins (the Thompson signing is market-implied at +0.85, but production-basis is unavailable, so the conservative entry is zero). The coach decision delta is also zero — Schottenheimer returns, same tendencies, same modeled gap. Add those together and the structural projection for 2026 is 7.4 wins.

If you want to be optimistic, you can layer in the market-implied value of Thompson (+0.85) and hope the rookie class contributes ahead of schedule. The ceiling read with those assumptions nudges toward 8.5. The floor, if Gary underperforms at 29 and the defensive rookies need development time, is probably 6.5. The central estimate is 7.4.

The falsifier is straightforward: if Dallas wins ten games, this read was wrong — the defense improved faster than the structural numbers projected, and the offseason additions delivered above their expected value. If the Cowboys win five, the read was also wrong, in the other direction — either Prescott's injury history caught up, or the cap concentration created roster gaps that weren't visible in the ledger. Anything between six and nine wins, and the model did its job.

The brief doesn't log any vacated opportunity — no major target-share departures left the room this offseason. That matters for fantasy construction because it means the Pickens addition doesn't come with a freshly cleared table; he's stepping into a room that already has CeeDee Lamb at one end of it. Two receivers in the top-six cap hits on the same team is an unusual configuration, and the production split between them is the central fantasy question for this roster. The brief doesn't resolve that split, and inventing a number wouldn't help you.

The sleeper case, if there is one, sits with Javonte Williams. Seventy-fourth percentile production at 57th percentile pay, 1.2% of the cap — those are the exact conditions the efficiency argument requires. He's not accumulating targets the way a receiver does, but as a volume back in a pass-heavy offense, the per-touch efficiency is there and the price hasn't moved to reflect it. Whether he sees enough volume to matter in standard formats depends on the scheme Schottenheimer runs, and that's a real uncertainty worth watching in the preseason.

No aging flags are logged in the brief for the skill-position core. Gary at 29 on a new edge contract is worth monitoring from a fantasy defense perspective — if the pass rush improves, opposing quarterbacks throw faster and shallower — but that's a team-level effect rather than an individual fantasy target. The data doesn't give us enough to build a fade case on any specific player here, and building one without the numbers would be filling a gap rather than reading the brief.

Everything above traces to a live instrument, and the receipts stay public: the GM boards carry Brian Schottenheimer's decision card, the team's cap posture ($249.9M already charged to future caps), and every contract verdict with its falsifier and review date — including George Pickens (FAIR PRICE), Rashan Gary (WATCH). When the season grades these reads, the grades post whether they flatter us or not.