Denver's 14-3 Record Was a Magic Trick. Here's the Math That Proves It.

The number that matters most from Denver's 2025 season isn't 14. It's 0.177. That's the gap between what the Broncos actually did — win 82.4 percent of their games — and what their point differential said they deserved to win, which was roughly 65 percent. Translate that back into a 17-game regular season and you're looking at a team that outperformed its scoring margin by somewhere between two and three wins. The 14-3 record was real. The foundation underneath it was considerably shakier.

This is the Pythagorean regression problem, and the NFL has a long history of punishing teams that ignore it. The formula, borrowed from baseball and adapted for football, uses a team's points scored and points allowed to estimate true winning percentage. Denver posted a plus-5.29 scoring margin per game last season — a legitimate number, not a mirage — but that margin predicted a good team, not a historically dominant one. The Broncos won close games at a rate that outstripped what their actual scoring edge would project. That's not skill, or at least it's not entirely skill. A significant portion of it is variance, and variance has a way of evening out.

What makes this more than an abstract statistical complaint is what it means for roster construction heading into 2027. The Broncos have real weapons. Bo Nix threw for 4,210 yards with 28 touchdowns in his first full season as the starter, and the receiving corps around him — Courtland Sutton's 1,087 yards and 8 touchdowns, Troy Franklin's 709 yards and 6 scores — is legitimate. J.K. Dobbins was extended this offseason, and the offensive core is intact. But the defense that helped manufacture those close wins is thin in ways the depth chart makes uncomfortable to look at. There's no viable depth behind Brandon Jones at safety. Zach Allen is the only defensive end with meaningful observed playing time, and he's on an expiring contract. Alex Singleton is 32 and starting at linebacker. These aren't opinions — they're structural facts about the roster.

The close-game problem compounds when you layer in the depth issues. Teams that win tight games at an unsustainable rate tend to regress toward their point differential the following season. When that regression arrives, it usually arrives in the form of a few extra losses — games that went your way last year that don't this year. If those losses come with a thinned-out defensive rotation and a receiving corps where the drop from Sutton to the second option is nearly 200 production units, the margin for error shrinks fast. Sutton is also on an expiring contract, which is a separate but related problem that the front office hasn't publicly resolved.

There's one more variable worth naming: Jonathon Cooper's situation. The outside linebacker was excused from mandatory minicamp following arrests on multiple charges across a seven-day span. Whatever the disciplinary outcome, it introduces more uncertainty into a linebacker group that already carries only two players with observed playing time from last season. My read is that the Broncos can absorb one disruption. They cannot easily absorb disruption stacked on top of structural thinness stacked on top of a schedule that won't keep gifting them one-score games. The 14-3 banner is hanging in Denver. The math says it was earned with a little help from variance. The question for 2027 is whether the roster is built to win without it.