Arbitrary Statistical Amalgam: Patriots To Win Super Bowl

As promised, procrastination notwithstanding, I’m back to look at the NFL playoffs from a somewhat statistical standpoint. Now, let me get this out of the way: while I guess I have the time to throw together some wacky new measurement to try to determine who will win the Super Bowl, I don’t really have the desire to figure out exactly what that measurement is to entail. So what I did was, prior to the playoffs, I busted out Excel and inserted into a spreadsheet a list of about three dozen stats and measurements from the 2007 regular season, though some categories also covered time prior to the ’07 campaign. The measurements ranged from things like the simple overall season rankings based on Football Outsiders’ signature metric, DVOA (as well as my own personal spin on the numbers FO provides with regard to offensive and defensive production) to coach’s winning percentage, to yardage per punt return, to which team won the most recent meeting between the two franchises.

At the end, I tallied up the “votes” in each category and came out with a projected playoff path provided by each measurable.

What I ultimately found was that — get this — based on as wide-ranging a statistical evaluation as I could come up with, the undefeated New England Patriots are the favorites to win Super Bowl XLII.

In fact, according to the original set of numbers, the Pats are nearly four times as likely as the next highest ranked team — the Indianapolis Colts — to end the season as the champions of the NFL.

For the record, the statistics covered such a broad range that of the 12 teams in the post-season tournament, 11 of them arose as possibilities to win the Super Bowl — only the Tennessee Titans failed to register as a potential Super Bowl victor.

Obviously, that means that three teams are no longer possibilities. Since I am trying to figure out a trend for NFL post-season success outside of tired clichés like “defense wins championships!”, I decided to remove any measurement that contained incorrect Wild Card Round selections. Humorously (in the “oh look, I’ve failed again!” sense), this left me with only six of my original 36 categories. Before I reveal those six, let’s take a look at the original Super Bowl percentages of likely Super Bowl victory this experiment produced. I should note that I’m aware that this adds up to a little more than 100 percent, but it’s just easier to round up sometimes.

  • New England 41.7
  • Indianapolis 11.1
  • Green Bay 8.3
  • Pittsburgh 8.3
  • Dallas 4.2
  • New York 4.2
  • San Diego 4.2
  • Washington 4.2
  • Jacksonville 2.8
  • Seattle 2.8
  • Tampa Bay 2.8

I almost listed Tennessee with a goose egg, but thought better of it.

Anyway, had I not spoiled the number of categories I had to remove earlier, it would have become apparent here that at least 15.3 percent of the statistics I had put my faith in had failed me. It’s glaringly obvious that this would greatly skew the numbers put forth in the original.

Of course, in comparison to the actual percentage of statistics that had to be removed (only 16.7 percent of the categories remain, in fact), 15.3 percent doesn’t seem that bad. Still, skewing is skewing, and we’ll have none of that regardless of its size (rounding up notwithstanding).

With New England’s dominance obvious in the first set, my initial thought — as someone who harbors great disdain towards Tom Brady — was that the revised, smaller sample might bring the Pats back down to Earth. Naturally, as had been the case all year when I thought Team Belichick would be knocked down a peg, I was in for more disappointment.

Indeed, of the six categories which accurately predicted victory for all four teams on Wild Card Weekend — points for, total offense, yards per punt return, team sacks, post-regular season playoff odds and recent history/wins in previous meetings — the Patriots are favored in four. In the case of the points for and playoff odds categories, the Pats would beat the Cowboys in Super Bowl XLII. Meanwhile, the Green Bay Packers would find themselves on the losing end of the Super Bowl should total offense prove to have significant predictive value. And, in a bittersweet accomplishment for the favorite team of yours truly, the Seahawks would be expected to fall to the Pats should recent history continue to pan out as it did in the wild card round.

My only hopes, from a Pats-hating standpoint, are that games are heavily influenced by punt returns (advantage: Seahawks) or quarterback sacks (which would work in the favor of the New York Giants.) For symmetry, here are the updated “odds”:

  • New England 66.7
  • New York 16.7
  • Seattle 16.7

Interestingly (to me, anyway), of the remaining six categories, the only one which managed to accurately forecast the Super Bowl XLI champion was sacks — excellent news for the Giants. Additionally encouraging for the G-Men is the fact that in 2004, the Seattle Seahawks and Pittsburgh Steelers were projected to land in the Super Bowl based on the team sacks metric. Given my standing as a Seahawks fan, I’ll refrain from going into further detail when I note that, while Seattle failed to win the game as its overall sacks lead predicted, some feel they didn’t get a fair shake in Jerome Bettis’s final game. I should also note that the team the Giants are projected to face (and defeat) in the Super Bowl should the metric pan out is our old friend New England.

While a sample size of two is perilously small, it seems to be the most notable of the remaining metrics. Last year’s leader in total offense, New Orleans, fell in the NFC Championship game. The highest-scoring offense in the NFL in 2006, the San Diego Chargers, also the pre-playoff favorite according to oddsmakers, failed to get past the divisional round. The Baltimore Ravens were to be beneficiaries of recent meetings with playoff foes, but they, too, failed to reach the AFC Championship game. The best of the rest, so to speak, was the Chicago Bears, who led the NFL in yards per punt return thanks to Devin Hester. They, of course, reached the Super Bowl before bowing to the Colts.

What it all boils down to is that, no matter how many statistics are assembled, their predictive nature is lower in the NFL than in other leagues — “any given Sunday” and all that. Just as a sample size of two is perilously small (demonstrated by further examination of the predictive nature of sacks: the Falcons would’ve knocked off either the Colts or Patriots in ’04, the Ravens would’ve beaten the Giants in ’03 and the Eagles would’ve won in ’02), an increase to 16 does not seem all that significant for accumulating data relevant to expected performance.

So, while Patriots fans have reason to be excited about their team, a small sampling of history suggests that they should be weary of the thought that their team is destined to make history of its own.

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: