Quarterback Wins Are Not a Stat
February 5, 2026 · NFL · 6 min read
Nothing reveals the intellectual bankruptcy of a football argument faster than the phrase "but he's 42-11 as a starter." Congratulations. You've just told me this person plays on a good football team. That is the entirety of the information conveyed. You might as well tell me his jersey number and expect me to evaluate his arm strength. Quarterback win-loss records are the astrological signs of football analysis — deeply felt, widely cited, and correlated with absolutely nothing useful about the individual in question.
The Math Is Not Complicated
Football teams have 53 players on the roster. Games are decided by offense, defense, special teams, coaching, and a healthy dose of randomness (fumble bounces, tipped interceptions, weather). Attributing the binary outcome of this 60-minute, 22-player-at-a-time contest to a single individual is not analysis. It's a horoscope.
Let's quantify this. The correlation between quarterback EPA/play (a genuine individual performance metric) and team win percentage is approximately r = 0.55. That means the quarterback's individual performance explains roughly 30% of the variance in wins. The other 70%? Defense. Running game. Coaching. Special teams. That one inexplicable muffed punt in the third quarter.
Historical Embarrassments
Some fun examples of why QB wins are absurd:
| QB | Record | EPA/Play Rank | What This Tells You |
|---|---|---|---|
| Trent Dilfer (2000) | 11-1 (reg) | Bottom third | His defense was historically great |
| Dan Marino (1994) | 10-6 | Top 5 | His defense was historically bad in the playoffs |
| Jimmy Garoppolo (2019) | 13-3 | Average | His team ran the ball and played elite defense |
| Matthew Stafford (2019) | 0-8 | Top 10 before injury | His team was the Lions |
Trent Dilfer has a Super Bowl ring. Dan Marino doesn't. If your evaluation framework concludes that Trent Dilfer was the superior quarterback in any year, your framework is broken, and you should feel bad about it.
"Winning Is A Skill Though"
Is it? Show me the mechanism. Show me the repeatable, measurable ability that a quarterback possesses — independent of his supporting cast — that consistently produces wins beyond what his individual stats would predict.
I'll wait.
What you'll find instead is that "clutch gene" quarterbacks who outperform their stats one year tend to regress the next. Because the "skill" you're observing is actually "having a good defense" or "opponents' kickers missing field goals" or "recovering an unusual number of fumbles." These are things that happen to a quarterback, not things a quarterback does.
Evaluating QBs by wins is like evaluating a drummer by the band's album sales. Travis Barker isn't better than Neil Peart because Blink-182 outsold Rush. That's not how any of this works.
The Alternative
If you want to evaluate a quarterback, look at EPA/play, CPOE, PFF passing grade, and QBR. These metrics aren't perfect, but they at least attempt to measure what the quarterback actually did on individual plays, rather than what his defense, kicker, and punt returner did for 60 minutes.
Related: Hot Takes Debunked · NFL Coverage