John Harbaugh Inherits a Roster That Was Better Than Four Wins

The Giants went 4-13 last season, and that number is almost certainly lying to you. Run their point differential — minus-58 on the year — through the standard points-per-win conversion and you get a structural win total of 6.9. The actual record was 2.9 wins worse than what the scoring margin said they deserved. That gap, a luck delta of -2.9, is not a small rounding error. It's the difference between a team that looks like a rebuild and a team that looks like an unlucky six-win squad.

The one-score record tells you exactly where those wins went. The Giants were 1-7 in games decided by eight points or fewer. Close games are mostly noise — a few plays, a bounce, a missed field goal — and a 1-7 record in them is about as bad as it gets. Teams this far on the wrong side of close-game variance almost always correct toward the mean the following year, whether they change a single player or not.

That's the single most important offseason fact about this franchise. Brian Daboll's tenure ended with a team that was outscored by 58 points and lost nearly every coin flip. The more honest ledger says they were a 6.9-win team in a 4-13 wrapper. Whatever comes next starts from a better foundation than the standings suggest.

Almost nothing worked on offense, but almost nothing catastrophically failed either — which is its own kind of problem. The Giants posted an offensive pass EPA of +0.01 per attempt and a rush EPA of +0.03 per carry. Those are slightly positive numbers on a 4-13 team, which sounds encouraging until you remember that slightly positive passing offense and slightly positive rushing offense can coexist with a historically unproductive season if the volume isn't there and the situation work collapses. Their red-zone EPA per play was -0.26, which means they were actively losing expected points when they got close enough to score. That's where games get decided.

The defense had a real problem on the ground. Opponents generated +0.13 EPA per carry against them — a number that ranks among the league's worst front-seven exposures. Pass defense was a different story; the Giants held opponents to +0.04 EPA per passing attempt, which is workable. The defensive architecture, then, was a pass rush and coverage unit that could hold its own in the air backed by a run defense that could not hold much of anything.

On fourth down, Daboll's decision model graded out at a 73.8% agreement rate with our model — below average — and his in-game choices cost the Giants an estimated 0.565 expected wins over the season, about 0.49 per season on a career basis. He went for it on fourth down 26.2% of the time; our model would have gone 44.7%. That gap between actual aggression and optimal aggression was a quiet tax on every close game this team played.

The most important entry isn't a player. John Harbaugh replaces Daboll as head coach, bringing a career expected-wins-lost rate of 0.31 per season against Daboll's 0.49. That 0.17-win swing from better fourth-down and situational decisions is real, and it accrues before the roster runs a single play.

Among the player arrivals, Isaiah Likely comes over from Baltimore on a three-year deal averaging $13.3 million annually. The market is paying 97th-percentile tight-end money for a player whose 2025 production sat at the 53rd percentile. Our aging curves give him a 1.11 multiplier at 26 — meaning modest growth is plausible — but the exit math is unfriendly: $15 million in dead cap if this goes sideways before 2027. The verdict here is an overpay, and it's wrong only if Likely finishes as a top-5 tight end in 2026. Tremaine Edmunds (LB, three years, $12 million annually) and right tackle Jermaine Eluemunor (three years, $13 million annually) both earn a WATCH grade — priced at the 97th and 93rd percentiles for their positions respectively, with exits cheap enough ($3.7 million and $4 million in dead cap) that the Giants aren't trapped if either underperforms. Greg Newsome II at corner and D.J. Reader at interior defensive line round out the arrivals at more modest price points.

The departures that matter most: Wan'Dale Robinson leaves for Tennessee on a $17.5 million annual deal after producing 140 targets and 217.9 PPR points in 2025. Daniel Bellinger and Cordale Flott also headed to Tennessee. The draft class adds chart value at the top, with fifth-overall pick Arvell Reese (LB, Ohio State, 2.05 chart wins) and tenth-overall pick Francis Mauigoa (OT, Miami, 1.58 chart wins) anchoring a class that also includes corner Colton Hood in the second round (1.05 chart wins) and receiver Malachi Fields in the third (0.66 chart wins). Total class chart value is meaningful; translating chart wins to real wins takes longer.

The honest net is close to neutral. On a production basis, the offseason moves cost the Giants -0.26 wins — Robinson's departure carries real production weight that Likely's arrival doesn't fully offset on current numbers. On a market basis, the net is -0.09 wins. Neither number is alarming; neither is encouraging. The Giants didn't meaningfully improve or damage their roster through free agency, and the draft class won't contribute significantly in year one.

The cap posture deserves a look. The Giants are carrying $133.1 million in future proration — dead-cap obligations from prior deals that will constrain flexibility in coming years. Their biggest positional investments sit in edge rusher (Brian Burns at 7.1% of cap), cornerback (Paulson Adebo at 8.1%), and safety (Jevon Holland at 6.2%), and those three groups carry the league's highest z-scores on the roster — meaning the Giants are paying well above market concentration for those positions. The one genuine albatross is Darius Slayton, consuming 5.3% of the cap while producing at the 31st percentile. The one genuine windfall is Malik Nabers on his rookie deal — 96th-percentile production at 2.7% of cap. That's the kind of structural advantage a rebuild needs to last.

The aging curves don't flag anyone arriving as a high-risk fade, but they also don't suggest the Giants found a discount on ascending talent. This is a roster being reshaped around a new coaching staff rather than one that made a decisive jump in quality. The right read is cautious optimism: the infrastructure is more stable than the 2025 record implied, but the ledger hasn't added enough to move the structural number much on its own.

Here is the arithmetic, plainly: the 2025 point differential projects to a structural base of 6.9 wins. The offseason moves, on a production basis, subtract 0.26 wins. The coaching upgrade — Harbaugh deciding at his career historical rate versus Daboll's — adds back 0.17 wins. The sum is 6.8 structural wins for 2026.

That is an accounting projection from the ledger, not a pick, not a bet. It says: given what the scoring margin showed last year and what the offseason added and subtracted, a team in this configuration typically wins about 6.8 games. The luck correction from that 1-7 close-game record is already baked into the structural base — the Giants don't get extra credit for it, they just stop being penalized by it.

The falsifier is straightforward. If the Giants win five games or fewer, this read is wrong — it means the 2025 underlying performance was less real than the point differential suggested, or the roster degraded in ways the ledger didn't capture. If they win nine or more, the ledger undersold something significant, most likely Harbaugh's impact or Nabers' leap. Somewhere between seven and eight wins is where this projection lives, and nothing about that range should surprise anyone who watched the scoring margins last fall.

Wan'Dale Robinson's departure is the number that matters most for fantasy purposes. He drew 140 targets and accumulated 217.9 PPR points in 2025. Daniel Bellinger added 26 targets and 59.6 PPR points before also leaving for Tennessee. Combined, that's 166 targets and 277.5 PPR points walking out the door — a substantial vacancy in a passing offense that, at +0.01 EPA per attempt, was at least functional.

The most obvious beneficiary on the roster is Malik Nabers, whose usage should expand further into that vacancy. He was already operating at the 96th percentile of receiver production on 2.7% of the cap — the market genuinely hasn't caught up with what he does per opportunity. More targets into a proven efficiency profile is the simplest case in this offense.

The sleeper case belongs to Isaiah Likely, with a caveat. He produced 61.7 PPR points on 36 targets last season — a per-target rate of 1.71 — in a Ravens offense that rarely needed him as a primary option. In New York, with Robinson's 140 targets vacated and Bellinger's 26 gone as well, Likely steps into a meaningful role for the first time. The efficiency is real; the opportunity is now genuinely open. The contract price flagged him as an overpay for win-value purposes, but for fantasy purposes, the question is simply whether the targets arrive — and the departure math says they should. That's what a sleeper actually is: a player whose role expanded faster than the consensus noticed.

Everything above traces to a live instrument, and the receipts stay public: the GM boards carry John Harbaugh's decision card, the team's cap posture ($133.1M already charged to future caps), and every contract verdict with its falsifier and review date — including Isaiah Likely (OVERPAY), Jermaine Eluemunor (WATCH), Tremaine Edmunds (WATCH). When the season grades these reads, the grades post whether they flatter us or not.