Tuesday, August 29, 2017

2017 NFL Preview: Looking at Improvement vs. Regression


So I was starting my research for the 2017 NFL Over/Under Challenge - Get in on this - only $30! - when somebody on some podcast who's probably smarter than me (proof: I already forgot who) was talking about regression - they said there were two statistics that most often vary from season to season (meaning: hardest to repeat from year to year): Turnover differential and records in close games. I was intrigued.

As American hero, Dr. Steve Brule would say ... Lets check it out!

2015 Turnover Differential:

I went back to 2015's numbers to see if those teams with the best turnover differential did, in fact, regress in 2016. Here are the top 5 teams, with their win differential from 2015 to 2016:

  1. Carolina Panthers: -9 wins
  2. Kansas City Chiefs: +1 win
  3. Cincinnati Bengals: -6 wins
  4. Arizona Cardinals: -6 wins
  5. Tie: Seahawks (Even), Patriots (+2 wins), Giants (+5 wins)
Analysis: Obviously, this set of data is not full proof. However, I see teams like New England and Seattle as the gold standard of the league - the exceptions to the rule - and wouldn't necessarily apply regression theories to either.

Meanwhile, Kansas City ranked 2nd in TO differential in 2015 and T-1st in 2016, which makes sense given their risk averse QB and playmakers like Marcus Peters and Eric Berry on defense. Again, only studying the numbers isn't full proof. Logic still applies.

However, I also see the three most disappointing teams of 2016 on this list - Carolina, Cincinnati and Arizona. Noted. Lets move on...

Here are the bottom 6 teams in 2015, starting with the worst, along with their win differential from 2015 to 2016:
  1. Dallas Cowboys: +9 wins
  2. Baltimore Ravens: +3 wins
  3. Tennessee Titans: +6 wins
  4. Jacksonville Jaguars: -2 wins
  5. Cleveland Browns: -2 wins
  6. Atlanta Falcons: +3 wins
Analysis: Obviously, I included a 6th team because Atlanta took a giant leap in 2016 - along with Dallas, Tennessee and even Baltimore, I guess. Again, not full proof - sometimes teams stay on top or bottom for a reason (BLAKE BORTLES FLASHING NEON SIGN).

2015 Record in Close Games:

Back to the 2015 numbers to see if teams who won a lot of close games in 2015 regressed the next year. The Top 5 teams in terms of Win % in one score games in 2015, along with their win differential from 2015 to 2016:
  1. Carolina Panthers: -9 wins
  2. Arizona Cardinals: -6 wins
  3. Denver Broncos: -3 wins
  4. San Francisco 49ers: -3 wins
  5. Minnesota Vikings: -3 wins
Analysis: This data comes off more conclusively. Many of the disappointing teams from 2016 are on this list - most notably, Carolina and Arizona, who landed on both lists. 

Here are the bottom 5 teams in terms of Win % in one score games in 2015, starting with the worst, along with their win differential from 2015 to 2016:
  1. Cleveland Browns: -2 wins
  2. Tennessee Titans: +6 wins
  3. Dallas Cowboys: +9 wins
  4. New York Giants: +5 wins
  5. San Diego Chargers: +1 win
Analysis: Wow. So the regression theory, as it relates to close wins, was correct in 9/10 cases. Once again, the teams you see show up on both lists were the most extreme cases - Dallas and Tennessee here. You also get the Giants somewhat cancelling out their 2015 TO differential numbers with their record in close games.

So now we've studied the numbers, as they applied from 2015 into 2016. All bubble gum and candy beans. Lets now apply it to the 2016 numbers, heading into the 2017 season.

2016 Turnover Differential:

Top 5 teams...
  1. Oakland Raiders
  2. Kansas City Chiefs
  3. New England Patriots
  4. Atlanta Falcons
  5. Minnesota Vikings
Bottom 5, starting with the worst...
  1. Chicago Bears
  2. New York Jets
  3. Jacksonville Jaguars
  4. Cleveland Browns
  5. LA Rams
2016 Record in Close Games:

Maybe you looked at the 2015 numbers and don't trust TO differential, but are intrigued by the record in close games regression theory. Here are the top 5 teams in Win % in close games in 2016:
  1. Oakland Raiders
  2. Houston Texans
  3. Miami Dolphins
  4. New England Patriots
  5. Dallas Cowboys
And the bottom 5, starting with the worst...
  1. Philadelphia Eagles
  2. Chicago Bears
  3. San Francisco 49ers
  4. Cleveland Browns
  5. Cincinnati Bengals
Analysis: Teams that show up on both lists: Oakland and New England, in terms of regression, and Chicago and Cleveland, in terms of improvement. Now, logic is screaming at me not to bet against New England nor on Cleveland, which means Oakland and Chicago are the prime candidates here to regress and improve, respectively.

2017 Improvement vs. Regression Rankings

I've gone through and assigned numeric values to each team's ranking in both TO differential and record in close games. In theory, the higher the score - the more likely the team improves in 2017, the lower the score - the more likely the team regresses in 2017.
  1. Chicago - 63 - MOST LIKELY TO IMPROVE
  2. Cleveland - 58
  3. Jacksonville - 56
  4. San Francisco - 55
  5. LA Chargers - 54
  6. NY Jets - 51
  7. Carolina - 46
  8. LA Rams - 46
  9. New Orleans - 43
  10.  Arizona - 42
  11.  Indianapolis - 40
  12.  Cincinnati - 40
  13.  Philadelphia - 40
  14.  Denver - 36
  15.  Buffalo - 33
  16.  Detroit - 33
  17.  Washington - 31
  18.  Houston - 29
  19.  Tampa Bay - 29
  20.  NY Giants - 28
  21.  Tennessee - 28
  22.  Minnesota - 27
  23.  Baltimore - 27
  24.  Seattle - 27
  25.  Atlanta - 22
  26.  Pittsburgh - 19
  27.  Miami - 18
  28.  Green Bay - 16
  29.  Dallas - 16
  30.  Kansas City - 9
  31.  New England - 7
  32.  Oakland - 2 - MOST LIKELY TO REGRESS
Analysis: There's a lot of trash up top and a lot of gold at the bottom, which I guess is what you'd expect when looking at teams most likely to improve vs. regress.

Again, I reiterate, I wouldn't necessarily apply this analysis to the stalwarts (the exceptions to the rule) such as New England, Seattle, and Green Bay and on the other end: Cleveland, Jacksonville and the NY Jets. Although, deciding which teams qualify for this can be tricky: Is Chicago a doormat now? San Francisco? Is Pittsburgh officially a powerhouse? Green Bay? etc.

Once you get past those teams, you see the strong candidates emerge. Granted, we've mentioned Oakland and Chicago as the prime candidates for each category.

***The teams who are primed for improvement in 2017: Chicago, LA Chargers, Carolina, LA Rams, New Orleans and Arizona (I'd also throw in Cincinnati and Philadelphia, personally).

***And the teams who look apt to regress in 2017: Oakland, Kansas City, Dallas, Miami and Atlanta (with nods to Houston and the NY Giants).

We'll see.


I'm a nerd. I thoroughly enjoyed gathering all this data. And you should put this article to some damn use and take part in the 2017 NFL Over/Under Challenge - A contest with nearly $1,500 in the pot in its inaugural season and a great way to follow all 32 teams all season long.

His Dorkness

No comments: