I am endeavoring to construct a predictive model like every other basement-dwelling football nerd in the United States, and the current one seems to be working competently. My model is based solely upon numbers and therefore cannot take into account intangibles such as key injuries, emotion, outside factors, or stupid coaching decisions (more on this in a moment). I first constructed the model prior to the Alabama-Notre Dame BCS championship game in 2013. The college model works differently than the pro model because the statistics have to be recalculated with the garbage games (e.g. FCS opponents and Kansas if you're in the Big Twelve) removed to give a clearer picture. That model pointed to an Alabama blowout of Notre Dame, which is precisely what happened. I used the model again for the Ohio State-Alabama semi-final Sugar Bowl game, and unlike the pundits on television, the model pointed towards a toss-up that favored Alabama (Ohio State won the game primarily due to Alabama's inability to stop the Buckeye offense). I first used this model last year for the Super Bowl, and it pointed towards a Seattle win over New England, 28-24. Keep in mind that although the score was correct but the winner wrong, no statistical model can calculate or take into account variables like a coach calling an insanely stupid play with the game on the line. Basically, it predicted a 4-point Seattle win and had they done what they should have (hint: handoff to Lynch), the final score of Seattle by three would have certainly been close. I used the model again for the NCAA football playoff, and it predicted an Alabama win over Michigan State by a count of 31-10. So certain was I that I'd be accused of rigging the model in favor of my own team that I added a touchdown for Michigan State and predicted a 14-point Alabama win.
Perhaps I should have stuck with the model.
I then used it to determine that Alabama would beat Clemson by ten points in the national championship game. The Tide led by 12 with a minute remaining before giving up a garbage-time touchdown to Clemson, so the model is still vindicated. Hence, I'm going to use it for the Super Bowl.
WARNING: Do not go and bet on one team or the other simply because this model has worked before. No statistical model can allot for all of the factors. If Cam Newton gets hurt on the fifth play from scrimmage (this is also known as the Colt McCoy clause), all bets are off. But let's look at the overall evaluation and see if we can draw a rational conclusion.
OFFENSE
The first thing that stands out about Carolina, of course, is their offense as led by league MVP Cam Newton. The Panthers have the most lethal offense in the NFL, averaging 32.2 points per game against a schedule that won 44.1% of their games. And while it was common to bash Carolina as running up numbers against pedestrian competition, they have hit full throttle since November 8th. In the last ten games, Carolina has topped 33 points seven times, had another game where they scored 31, and had two contests where they were held below their average, a 27-point win over the Titans and an absolutely mind-boggling 13 points against the Atlanta Falcons in their only loss of the season. If you hold as I do that football is as simple run, stop the run, and win the turnover battle, Carolina is #2 in the league in rushing. At first glance, you wonder how this is even possible. Their leading rusher only obtained 989 yards in 13 games, but Cam Newton added another 636 yards on his own. Although Carolina is almost exactly 50/50 in run plays/pass plays, they get a lot more out of their running game despite a completion percentage of nearly 61% from the quarterback.
Denver's offense is more pass-oriented in play calling but more balanced in outcome. Denver passes the ball three of every five plays, and that average would probably be higher if Peyton Manning had not been injured for several games late in the year. How balanced? Denver has the #17 rushing offense, #17 passing offense, and is 18th in the league in offensive ppg with 22.1. They have played a minimally more difficult schedule than Carolina although such things mean very little in the parity of the NFL as opposed to the wide talent gap in college football. Carolina, in fact, has played seven rushing offenses better than Denver and beaten them all. They've played ten offenses that scored more ppg than Denver and beaten all of them as well. The flip side, of course, is that Denver's offense was 5.1 ppg worse when Manning was out than in games he played. Thus, one may reasonably conclude that Denver's offensive rating with Manning healthy in all sixteen games would (if extrapolated) lift their offense from 18th to 10th in ppg. On the other hand, Oswiler did score more points against the Patriots in his win than Manning did. In short, Manning is worth an extra touchdown plus experience.
Consider this: in all but one game this year, Carolina has scored or exceeded the average ppg surrendered by the opposing defense. The one exception? Their loss to Atlanta in week 15. Put another way, Carolina scores points regardless of how good the opposing defense is. The one defense Carolina faced better than Denver was Seattle, whom they played twice. And Carolina ripped the Seahawks defense that gave up an average of 17.6 ppg for 27 and 31 points (although I'll grant one touchdown was a defensive TD). Denver, on the other hand, has only exceeded the points allowed by the opposing defense nine times in eighteen contests.
One other observation is that Carolina has won games by virtually all means possible. They've blown teams out early, fallen behind by two touchdowns and roared back, and blown big leads and recovered just in time to win. Denver has been more like the resilient team that sticks with their game plan and adjusts as necessary.
DEFENSE
The first thing that grabs your attention when watching Denver is the speed of their defense, particularly their run defense. Denver, in fact, has the best rushing defense in the NFL, surrendering a paltry 81.4 yards per game. This means the game matches strength against strength. You would think that their passing defense would therefore be somewhat weaker, but if anything, their pass defense (3rd in the league) is just as good. The weakness of Carolina's defense, in fact, is the passing defense that surrenders 239.1 yards per game. But there's also a catch (if you'll pardon the pun) when you consider who has beaten Denver this year: Indianapolis, Kansas City, Oakland, and Pittsburgh. Only the Chiefs have a better passing defense than Carolina, meaning that three passing defenses worse than the Panthers have beaten the Broncos. Of course, that, too comes with a caveat: only Indianapolis beat Peyton Manning, and he was injured during that game. The other three losses were all under quarterback Brock Oswiler and Manning's defeat of Kansas City might be considered to offset their passing defense against Oswiler as well.
But all statistics contain both truth and misleading notions. Sure, Carolina's pass defense numbers are not overly impressive but then you remember that three of their sixteen games came against Drew Brees (two games totaling 574 yards) and Aaron Rodgers (another 331 yards). Subtract those totals from Carolina's year and the Panthers move from 16th to 5th in passing defense. And then remember that despite all those passing yards, the Panthers still won all three games. Furthermore, how many of those yards are because teams are so far behind early that they're reduced to one-dimensional passing?
Denver's fourth-ranked overall defense allows 18.3 ppg while Carolina's sixth-ranked defense allows 19.3 ppg. Denver held 14 of their 18 opponents this year below their normal average offensive output while Carolina did so in 13 contests. Denver's defense gets a slight edge because of the stronger competition.
TURNOVERS
While Carolina takes the offense (although closer with Manning playing) and Denver takes the defense, it is the turnovers that vastly separate the two teams in ways that boggle the mind. Neither team fumbles the ball much: Carolina averages less than one fumble per game and Denver averages one. Carolina averages 1.1 turnovers per game while Denver average 1.8. The hard statistic that favors Carolina is that Denver throws twice as many interceptions per contest as Carolina does, and Manning is particularly susceptible to the pick.
But if there is one eye-popping statistic buried beneath all of the numbers, it is this: Carolina averages 9.25 points per game from turnovers. In other words, if Denver played only offense the entire game and Cam Newton never took the field except via a turnover, Carolina would on average score nine points solely due to their defense forcing a turnover and capitalizing. Denver, on the other hand, would score 4.9 ppg via the turnover but give up 6.25 per game - a minus of 1.35 points. Even when Carolina loses the ball via turnover, they only surrender an average of two points per game. Thus, Carolina is at +7.25 while Denver is at -1.35. This fact alone means that all things being equal, Carolina starts the game with an average advantage of 8.6 points.
PENALTIES
Carolina gets flagged for 6.2 penalties per game and loses 53 yards via the flag. Denver gets flagged for 7 penalties per game and loses 64 yards. This is a net of two first downs in favor of Carolina, one via the yardage difference and one via the number.
PREDICTION
Most of the numbers favor Carolina and most likely in a game that gets out of hand. If you take the mean of the average Carolina offensive ppg and the Denver defense and allow for a slight difference in strength of schedule but an even greater difference in expected output, Carolina is very likely to exceed the 18.3 ppg Denver normally surrenders and probably by a wide margin. Denver, on the other hand, is likely to lose the turnover battle and be fortunate to get to the 21 points they would average. The one positive in Denver's favor (aside from their stout defense) is that their passing offense combined with Carolina's second-half woes throughout the season mean that even a huge halftime lead is not an automatic win for the Panthers. My prediction based on the numbers is listed in the 'expected range,' and I'm picking Carolina by eight. The statistical model suggests:
EXTREME RANGE: Carolina 33 Denver 14
EXPECTED RANGE: Carolina 28 Denver 20
UPSET RANGE: Denver 27 Carolina 24
CAVEATS
As noted earlier, intangibles are part of the game. Maybe Peyton Manning comes out with a little extra fire in what figures to be his last game in the NFL. Maybe Cam Newton freezes on the big stage (given his prior accomplishments, this would really surprise me). Maybe Denver gets two quick turnovers for two quick scores and holds on in a close game. A prediction is just that - a prediction and nothing more. So don't email me about what an idiot I am if Denver does happen to win, particularly since I will be pulling for the Broncos anyway. But this is what the numbers suggest.
