≡ Menu

Here is a recap of the 2012 Chicago Bears season.  Notice anything strange? Trick question!

Week Day Date OT Rec Opp Tm Opp
1 Sun September 9 boxscore W 1-0 Indianapolis Colts 41 21
2 Thu September 13 boxscore L 1-1 @ Green Bay Packers 10 23
3 Sun September 23 boxscore W 2-1 St. Louis Rams 23 6
4 Mon October 1 boxscore W 3-1 @ Dallas Cowboys 34 18
5 Sun October 7 boxscore W 4-1 @ Jacksonville Jaguars 41 3
6 Bye Week
7 Mon October 22 boxscore W 5-1 Detroit Lions 13 7
8 Sun October 28 boxscore W 6-1 Carolina Panthers 23 22
9 Sun November 4 boxscore W 7-1 @ Tennessee Titans 51 20
10 Sun November 11 boxscore L 7-2 Houston Texans 6 13
11 Mon November 19 boxscore L 7-3 @ San Francisco 49ers 7 32
12 Sun November 25 boxscore W 8-3 Minnesota Vikings 28 10
13 Sun December 2 boxscore L OT 8-4 Seattle Seahawks 17 23
14 Sun December 9 boxscore L 8-5 @ Minnesota Vikings 14 21
15 Sun December 16 boxscore L 8-6 Green Bay Packers 13 21
16 Sun December 23 boxscore W 9-6 @ Arizona Cardinals 28 13
17 Sun December 30 boxscore W 10-6 @ Detroit Lions 26 24

The 2012 Bears played two terrible teams, the Titans and the Jaguars. Those were the two biggest blowouts of the season for Chicago. The Bears had five games against really good teams (Seattle, San Francisco, Houston, and the Packers twice): those were the five biggest losses of the season. Chicago had one other loss, which came on the road against the next best team the Bears played, Minnesota.

I see some strange

I see some strange.

All of that might sound …. unsurprising to you. And that’s the point. Six years ago, Doug wrote a post titled strange seasons, after he noticed an odd quirk about the 2006 Jaguars. That year — and without adjusting for strength of schedule — Jacksonville played much better against good teams than it did against bad teams.

To measure how “strange” a season was, Doug measured the correlation coefficient between (a) the SRS rating of a team’s opponent in each game, and (b) the margin of victory in that game. You would expect a negative correlation generally: that is, as the rating of the opponent increases, the margin of victory over them should decrease. The 2007 Jaguars had a positive correlation, which is strange, which was the origin of Doug’s post.

I decided to run “strangeness” correlation coefficients for each team in each 16-game season in NFL history, using SRS rating and location-adjusted margin of victory (i.e., awarding three points to the home team) as my two variables. As it turns out, all 32 teams in 2012 had negative correlation coefficients for those two variables (which is what you would expect). The table below lists (for reference purposes) each team’s average MOV, average SOS, and SRS rating in 2012, and the final column shows the CC between each team’s average opponent rating and average margin of victory (after adjusting for location). As you can see, the Bears had the “least strange season” while the Steelers had the “strangest season.”

Chicago Bears6.10.86.9-0.89
Buffalo Bills-5.7-1-6.7-0.75
San Diego Chargers0-2.3-2.3-0.74
Oakland Raiders-9.6-1.3-10.8-0.71
Tennessee Titans-8.8-1.2-10-0.67
Cleveland Browns-4.1-1.2-5.3-0.66
Denver Broncos12-1.910.1-0.65
Dallas Cowboys-
Washington Redskins30.43.4-0.62
Cincinnati Bengals4.4-2.42.1-0.6
San Francisco 49ers7.82.510.2-0.57
Indianapolis Colts-1.9-2.8-4.7-0.56
New York Jets-5.90-5.9-0.55
Houston Texans5.3-1.83.5-0.53
Arizona Cardinals-6.73.5-3.2-0.51
New Orleans Saints0.411.4-0.49
Green Bay Packers6.11.27.3-0.47
Baltimore Ravens3.4-0.52.9-0.43
Philadelphia Eagles-10.31.4-8.9-0.42
Detroit Lions-4.11.8-2.3-0.41
Miami Dolphins-1.8-0.8-2.6-0.4
St. Louis Rams-
Carolina Panthers-
Seattle Seahawks10.41.812.2-0.35
Tampa Bay Buccaneers-0.30.30-0.35
New England Patriots14.1-1.412.8-0.34
Jacksonville Jaguars-11.8-1.1-13-0.3
Minnesota Vikings1.91.43.4-0.29
Kansas City Chiefs-13.4-0.6-14-0.14
New York Giants5.30.96.2-0.13
Atlanta Falcons7.5-1.16.4-0.12
Pittsburgh Steelers1.4-2-0.7-0.05

But the Bears didn’t just have a predictable season. That -0.89 correlation coefficient is the lowest for any 16-game season in NFL history. In other words, Chicago just had the least strange season of the modern era. Here’s a game-by-game breakdown, like at the top of this post, but this time, I’ve sorted the schedule from toughest to easiest opponent. As you can see (and what the -0.89 means), there is almost a perfect negative correlation between the quality of the opponent and the (location-adjusted) final margin:


On some level, this isn’t that interesting. After all, it’s not a story when a team does poorly against great teams, pretty well against mediocre teams, and great against bad teams. But the fact that Chicago was so consistent in its performance is pretty interesting, and by ranking as the least strange season in over 30 years is very interesting. At least to me.

At the end of his post, Doug noted that there may be some predictive utility in this. I decided to run similar tests to see if it means anything to have a predictable or strange season. One could argue that the Steelers — who have shown an ability to play well against better teams, even if they slip up against lousy teams — are more likely to become a great team in 2013 than the Bears, because beating good teams shows a higher ceiling. That’s just a theory, though.

I ran a regression using Year N Wins and Year N Strangeness Coefficient as my inputs; the best fit formula was:

Year N+1 Wins = 5.45 + 0.360 * Year N Wins + 0.74 * Year N SC

The R^2 was 0.36 and the p-value on the Strangeness Coefficient was 0.107, on the border of being statistically significant. This formula would project the 10-6 Bears to win 8.4 games in 2013, while the 8-8 Steelers are projected to win 8.3 games. That’s pretty interesting and would support the “high ceiling” theory. But it’s far from obvious whether the strangeness coefficient has any predictive power. One reason is that the p-value is not very convincing. But here’s another. I re-ran the regression using Year N SRS Rating instead of Year N wins.

Year N+1 Wins = 8.33 + 0.20 * Year N SRS + 0.50 * Year N SC

The R^2 was 0.39 (this makes sense, since SRS is a better predictor than wins), but the p-value on the SC was just 0.27. I suppose the evidence indicates that, all else being equal, it’s better for an 8-8 team to have a really low or even positive Strangeness Coefficient than a really high one like the Bears. But I’m not sure if I’d say it’s anything more than a tiebreaker.

Putting aside all the talk of strangeness and regression and correlations, I think one conclusion is pretty obvious. Bears management saw how that Chicago was very predictable against the league’s best teams. Despite a 10-6 record, Chicago’s struggles against the upper crust of the NFC ultimately cost Lovie Smith his job. And that’s not very strange, either.

Previous “Random Perspective On” Articles:
AFC East: Buffalo Bills, Miami Dolphins, New England Patriots, New York Jets
AFC North: Baltimore Ravens, Cincinnati Bengals, Cleveland Browns, Pittsburgh Steelers
AFC South: Houston Texans, Indianapolis Colts, Jacksonville Jaguars, Tennessee Titans
AFC West: Denver Broncos, Kansas City Chiefs, Oakland Raiders, San Diego Chargers
NFC East: Dallas Cowboys, New York Giants, Philadelphia Eagles, Washington Redskins
NFC North: Chicago Bears, Detroit Lions, Green Bay Packers, Minnesota Vikings
NFC South: Atlanta Falcons, Carolina Panthers, New Orleans Saints, Tampa Bay Buccaneers
NFC West: Arizona Cardinals, San Francisco 49ers, Seattle Seahawks, St. Louis Rams

  • Danish

    Very interesting. I don’t now what to make of it, but interesting for sure. It feels like this should line up well with FOs VAR measure, but a quick eyeballing suggests otherwise – am I missing something or is SRS and DVOA just that different (which wouldn’t be that … strange)?

    The last couple of years Atlanta has felt like a team that was good overall, took care of business against the cupcakes but struggled against quality opponents and thus no threat to make a run at the Super Bowl. I had the same feeling about them last year, but this seems to show that I was way off with that… (They were 17th in VAR in 2012, after 3rd and 1st in 2010 and 2011 so maybe it was the force of habbit that led me astray. Do you have these rankings for 2011 and 2010 for comparison?)

  • xmenehune

    what if you used your predicting formula on the results of 2011 season and compared to 2012 season actual W-L records. What do you see?

    go back even farther say ’02 and see if there are a percentage of teams that are within a win or two of formula results. just wondering….

    • xmenehune

      also just thinking since there is a lot of turnover of personnel and staff each season, if the formula is close then perhaps it’s just a coincidence. Perhaps formula might be only useful for teams that had the least amount of turnover…

  • Scott Tanner

    Not going to disagree with the actual analysis. But trust me, as a Bears fan, a season that started 7-1 and then slowly devolved into an injury fueled nightmarish playoff-missing soul-annihilating collapse is not something I would first describe as “least strange”.