Skip to main content
The Carolina Panthers Fought The Stats — And Won

It may be a distant memory now, in the wake of a 17-1 record and an NFC championship, but at midseason the Carolina Panthers were a litmus test for the depth of a football fan’s sabermetric beliefs. Sure, the Panthers were undefeated, and they looked like a solid team. But there was nothing about their underlying numbers that made them seem like a steamroller hellbent for the Bay Area come early February. If anything, their impressive record seemed liable to regress toward the mean as the season went on.

Instead of backsliding toward their “fundamentals,” like many fast-starting teams have over the years, the Panthers did something remarkable: They actually improved their core metrics to match their record, and kept right on winning. Since midseason, Carolina truly has played like the best team in football, culminating in some of the most impressive early blowout wins in playoff history.

More Sports

How rare was Carolina’s victory over the fundamentals? Since 1978 (the advent of the 16-game schedule), a shade under half of all NFL teams have improved their schedule-adjusted1 Pythagorean winning percentage between the first and second halves of the schedule. Of those teams, a little more than half again saw the improvement move in the direction of the team’s midseason winning percentage. In other words, their actual record was better than their Pythagorean percentage at mid-year, and their Pythagorean percentage over the rest of the season improved toward their record. According to rest-of-season Pythagorean percentage, the 2015 Panthers finished as the 11th-best team of that group, which also includes the 1985 Bears, 1984 49ers and 2014 Seahawks — pretty heady company.

But what makes the Panthers’ improvement even more special is the degree to which their actual and Pythagorean records differed halfway through the season. At midyear, they were 8-0 but had the point differential of a 5.3-win team — the second-biggest disagreement of any team in our sample.2 When a team’s actual and Pythagorean records disagree that much, the Pythagorean number is right more often than not,3 though to be fair it’s close to a coin flip. But in Carolina’s case, their winning percentage was right on the money — only five other teams since 1978 saw their second-half Pythagorean improvement predicted more accurately by their first-half winning percentage. And all of those other cases featured teams floating around .500 with subpar point differentials — Carolina is the only team in modern history to post a stellar first-half record with solid fundamentals, then turn into a Pythagorean behemoth down the stretch.

Without a doubt, it was offensive improvement that drove Carolina’s overall uptick in performance, and practically all of that change has come in the passing game. According to expected points, the Panthers’ special teams only marginally improved in the second half, and their defense, while great, stayed pretty much the same all season.

Back in November, when I infamously described the unbeaten Panthers as the worst team ever to start 11-0, I also wrote that Cam Newton was having a “decent, but not great” season, and that “quarterback play probably isn’t the main driving force behind the Panthers’ success.” That may have been true at the time, but it surely isn’t the case now. Since midseason, Newton has been the second-most valuable QB in football according to combined passing and rushing value over average,4 and his rate statistics have undergone a particularly remarkable metamorphosis.

At midseason, Newton’s numbers looked much the same as in years past: His rates of completions, interceptions and sacks were subpar, and his yards and touchdowns per attempt were only average. He had one of the league’s highest rates of off-target throws5, with the only mitigating factor being that he also ranked among the game’s most frequent deep passers. (Deep passes are more valuable but harder to complete; Newton was inaccurate even if we adjust for this.) The majority of his value added was with his legs, as the game’s most prolific running QB. It was pretty much the same Cam we’ve seen the last few years.

But to say Newton has improved his passing since then is like saying he only slightly enjoys dancing in everyone’s mug. Since Week 10, he suddenly ranks among the league’s most accurate passers, with a drastically reduced rate of off-target throws. He hasn’t cut back on the deep tosses, either — his air yards per attempt have dipped only marginally in the second half of the season. He’s slinging the ball for significantly more yards and touchdowns per attempt, and he’s hardly ever giving it away; his TD-INT ratio is 21-1 over his last 8 games.

The changes to Newton’s game have come with a few tradeoffs: His sack rate is up and his rushing output slightly down since his midseason transformation began. But his value added per game has increased by a factor of nearly six over the season’s back stretch. That gives him the sixth-biggest second-half improvement of any Super Bowl starting QB since 1978. Few QBs in history have gone into a Super Bowl on a better roll than Newton is on right now.

Now, quarterback play is unquestionably the main driving force behind the Panthers’ success. The only remaining question is whether Newton can keep up his newfound passing production against one of the toughest defenses he’s ever faced. Given the remarkable way he and the Panthers overcame the odds and molded their fundamentals into those of a championship squad, I wouldn’t bet against them finishing the job on Sunday.

Check out our live coverage of Super Bowl 50.

A Super Bowl statistics special from our sports podcast Hot Takedown. Listen above, or subscribe on iTunes.


  1. Using our Elo ratings to measure the difficulty of each opponent at the time of the game.
  2. The 2005 Packers were 1-8 with the point differential of a 4.3-win team midway through the season.
  3. Specifically, among the upper quartile of “disagreements,” Pythagorean record is the better predictor of rest-of-season form 52 percent of the time after properly regressing first-half winning percentage and Pythagorean percentage to the mean for predictive purposes.
  4. Trailing only Seattle’s Russell Wilson.
  5. Pass attempts judged by ESPN’s Stats and Information group to have fallen incomplete only because Newton missed his intended target, whether because of an overthrow or underthrow.

Neil Paine is a senior sportswriter for FiveThirtyEight.

Filed under , , ,

Comments Add Comment