LONG RUN IN 2025
The Eagles finished dead last in third-down conversion rate in the second half of 2025 games — 29.2% third-down punt rate, worst in the NFL. Part of that was execution. Part of it was Jalen Hurts' inconsistency in the intermediate passing game. But a significant, quantifiable, documented part of it was play-calling. Specifically: running the ball on third-and-long.
We catalogued every Eagles rushing attempt on 3rd-and-5 or longer in the 2025 regular season. The results aren't just bad. They reveal a coordinator who was either ignoring data available on the sideline or overriding it for reasons the numbers cannot explain.
Here's what's important to understand about 3rd-and-long defenses: they are not stacking the box. They know a pass is coming. Safeties are playing deep halves or quarters. Linebackers are dropping into zones. Cornerbacks are in off coverage or pressed at the line. The run fit is actually softer than normal — which means a scramble or designed run can occasionally catch them.
That is not why the Eagles failed. They failed because no alignment makes a running back averaging 4.1 yards per carry suddenly gain 8, 10, or 13 yards on a single play. The defense isn't the constraint. The geometry of the situation is the constraint. GOPP measures whether the call matches what the game script demands. On 3rd-and-5 or longer, the game script demands a pass on approximately 94–96% of snaps across the league. The Eagles called pass on roughly 78%.
That 16-point GOPP gap means roughly 1-in-5 times the Eagles were in a pure passing situation, they handed the ball off instead. Not because the defense was soft against the run. Because the coordinator chose a call with a mathematically near-zero conversion ceiling — and did it 32 times.
EPA+ on a 3rd-and-long run is almost always negative — not because the defense is loading the box, but because the required gain exceeds what running plays produce at any reasonable rate. Barkley averaged 4.1 yards per carry in 2025. On 3rd-and-long he averaged 2.1 — but even if he'd hit his season average every time, a 4.1-yard gain on 3rd-and-8 is still a punt. The math fails before the snap.
Even in the best case — a defense in soft pass coverage with a gap open — Barkley would need to nearly double his season average on a single carry to convert a typical 3rd-and-long. That's not a reasonable ask. The EPA+ consequence is severe: every failed 3rd-and-long run that produces a punt generates approximately -0.7 to -1.1 EPA, because it hands the opponent the ball with full field position advantage and consumes a possession.
The correct framing: This isn't about predictability. The defense being in pass coverage doesn't help you if your runner can't gain 8 yards. GOPP says you should be passing. EPA+ says every time you don't, you're burning expected points. The 6.3% conversion rate is the mathematical outcome of those two facts colliding 32 times.
| Game / Situation | Call | Result | EPA+ Impact | GOPP Grade |
|---|---|---|---|---|
| vs 49ers (WC) · Q3 · 2nd & 18 Trailing 17–13 after Saleh adjustments |
Barkley run | -1 yard | -0.8 | F |
| vs 49ers (WC) · Q3 · 3rd & 18 Result of previous Barkley run on 2nd |
Goedert short (5 yds) | Punt | -0.7 | F |
| vs 49ers (WC) · Q3 · 3rd & 13 Later possession, still trailing |
Hurts scramble | 3 yards, punt | -0.6 | D |
| vs Bills (Wk 17) · 2nd half Eagles lead 13–0, protecting clock |
Barkley run on 3rd & 8 | 2 yards, punt | -0.5 | D- |
| vs Bears (Wk 12) · 3rd & 13 Down 24–7, must have conversion |
Shipley run | 1 yard, punt (the "WILL SHIPLEY???" game) | -1.1 | F |
The Yahoo Sports post-mortem on the Wild Card loss listed 12 instances of 3rd-and-long situations in the 49ers game alone. "A Barkley run on 2nd-and-18? A five-yard completion to Goedert on 3rd-and-18? Another Barkley run on 2nd-and-10 and a Hurts run on 3rd-and-13? A Barkley run on 2nd-and-20? WILL SHIPLEY? On 3rd and 13?" The frustration is understandable. The data makes it worse.
OPI doesn't actually indict the 3rd-and-long run specifically — defenses are in pass coverage, so the run isn't "predicted" in the traditional sense. What OPI does indict is the first and second down play-calling that kept putting the Eagles in 3rd-and-long to begin with. An OPI of 84 in the second half of games — the number Patullo posted against the 49ers — means defenses knew what was coming on early downs too. That's how you manufacture 3rd-and-10 situations: lose on 1st and 2nd down because the defense already has you dialed in.
The systemic failure is a two-layer problem. Layer one: high OPI on early downs created 3rd-and-long situations at an above-average rate. Layer two: once in those situations, the GOPP gap meant the Eagles were calling runs 16 percentage points more often than league average — on a play type with a 6.3% ceiling. One problem fed the other all season.
The 2026 correction: New OC Sean Mannion comes from the Shanahan tree (Green Bay). The Shanahan system creates favorable down-and-distance via run efficiency on 1st and 2nd down — which means fewer 3rd-and-longs to begin with. Fix the early-down OPI, and the 3rd-and-long run problem largely disappears on its own. Eaglytics will track Mannion's early-down OPI starting Week 1 of 2026. That's the leading indicator.
That's the real indictment. The GOPP gap of 16 percentage points means the Eagles were calling runs on 3rd-and-long roughly 1-in-5 times more than league average — on a play type with a mathematical conversion ceiling of approximately 6–8% even against a soft box. The EPA+ consequence of −0.9 per attempt is not about the defense. It's about the geometry of the situation.
And the OPI story is actually on first and second down — where Patullo's 84 OPI in the second half of the 49ers game told defenders exactly what was coming before the snap, manufactured negative plays, and put the Eagles in 3rd-and-long to begin with. Fix the early-down OPI and you don't need to solve the 3rd-and-long run problem. It solves itself.
Mannion's early-down OPI in Week 1 of 2026 is the most important number Eaglytics will publish all season. Watch for it.