You want to know the strength of your NFL team. You’ll take any analytics that can sort through the preseason noise of the NFL.
In college football, team strength tends to persist from year to year. This makes it possible to use previous seasons to predict the current season.
However, looking at past years does not work in the NFL since team performance regresses to the mean. The salary caps levels the playing field for all 32 teams. Injuries and luck can derail teams with the highest expectations, such as Atlanta in 2013.
However, we can use a different trick from college sports to rank NFL teams in the preseason. Let me explain.
Wisdom of many sports writers
Preseason polls in college sports are remarkable predictors of success.
I first learned about this counter intuitive result from Nate Silver, who uses the preseason AP college basketball poll in his NCAA tournament predictions.
The same accuracy holds for college football polls. In the preseason AP poll that gets released before the season, the higher ranked teams win 59.5% of bowl games that postseason, a result based on 300 bowl games since the 2005 season. The preseason Coaches poll has predicted a more remarkable 61.2% of bowl game winners in the same time span.
The combined wisdom of sports writers or coaches lead to remarkable rankings. However, the accuracy of these polls decrease once the season starts. The writers or coaches tend to react too strongly to wins and losses. By the end of the season, the higher ranked team in the AP polls wins 56% of bowl games.
However, the AP poll is a remarkable tool before the season starts. Let’s created the same type poll for the NFL.
Ensemble NFL preseason rankings
Every major sports media site publishes preseason power rankings. I looked at 20 from before the 2013 season.
A team’s rank isn’t enough to make game predictions. We also need a team’s rating, which gives an expected margin of victory over an average NFL team.
To do this, I took a team’s rank and assigned it a rating based on historical results from The Power Rank. For example, the top ranked team had a 9.7 rating from 2003 through 2012.
Now for each subjective power ranking, a team gets both a rank and a rating, just like the rankings here on The Power Rank. To get ensemble preseason rankings, a team’s rating is averaged over the 20 subjective power rankings. Here are the results for 2013 along with the team’s final record after the playoffs.
1. San Francisco, (14-5), 8.43.
2. Seattle, (16-3), 7.97.
3. Denver, (15-4), 7.35.
4. Atlanta, (4-12), 5.65.
5. Green Bay, (8-8-1), 5.29.
6. Baltimore, (8-8), 4.92.
7. New England, (13-5), 4.89.
8. Houston, (2-14), 4.38.
9. Cincinnati, (11-6), 3.04.
10. Washington, (3-13), 2.25.
11. New York Giants, (7-9), 1.58.
12. New Orleans, (12-6), 1.34.
13. Chicago, (8-8), 1.33.
14. Indianapolis, (12-6), 1.05.
15. Pittsburgh, (8-8), 0.18.
16. Dallas, (8-8), 0.15.
17. Minnesota, (5-10-1), 0.02.
18. St. Louis, (7-9), -0.79.
19. Carolina, (12-5), -1.35.
20. Miami, (8-8), -1.56.
21. Tampa Bay, (4-12), -1.95.
22. Detroit, (7-9), -1.99.
23. Kansas City, (11-6), -2.52.
24. Philadelphia, (10-7), -3.65.
25. Cleveland, (4-12), -4.11.
26. San Diego, (10-8), -4.20.
27. Tennessee, (7-9), -4.42.
28. Arizona, (10-6), -4.46.
29. Buffalo, (6-10), -5.11.
30. New York Jets, (8-8), -6.80.
31. Jacksonville, (4-12), -8.17.
32. Oakland, (4-12), -8.74.
Clearly, the preseason ensemble rankings thought too highly of Atlanta and Houston, two teams that combined for 6 wins in 2013. On the other end, the ensemble rankings missed low on San Diego and Arizona.
However, the rankings had the final four teams in the playoffs (Seattle, Denver, San Francisco and New England) in the top 10. In addition, they did predict 62.5% of game winners over the 2013 regular season and playoffs. The Vegas line gets 66% of games correct on average.
Preseason ensemble rankings for 2014
Here are results for the 2014 season.
1. Seattle, 9.61.
2. Denver, 7.96.
3. New Orleans, 6.58.
4. New England, 6.40.
5. San Francisco, 6.13.
6. Green Bay, 6.07.
7. Philadelphia, 3.72.
8. Indianapolis, 3.42.
9. Cincinnati, 3.10.
10. Chicago, 2.03.
11. San Diego, 1.85.
12. Arizona, 1.20.
13. Baltimore, 1.01.
14. Pittsburgh, 0.78.
15. Carolina, 0.39.
16. Kansas City, 0.30.
17. Detroit, -0.03.
18. Atlanta, -0.99.
19. St. Louis, -1.52.
20. Tampa Bay, -1.63.
21. Miami, -2.13.
22. New York Jets, -2.40.
23. New York Giants, -2.41.
24. Dallas, -3.44.
25. Washington, -3.78.
26. Minnesota, -3.95.
27. Tennessee, -4.60.
28. Houston, -5.56.
29. Jacksonville, -6.02.
30. Cleveland, -6.43.
31. Buffalo, -6.98.
32. Oakland, -8.69.
The rankings in the ensemble differed the most on Houston. The Texans have lots of question marks on offense with new QB Ryan Fitzpatrick. However, they had a decent defense last season and added top draft pick Jadaveon Clowney to a pass rush that already features J.J. Watt. Houston is 28th in the preseason rankings.
The rankings in the ensemble differed the least on Philadelphia. From all the Chip Kelly is a genius articles out there, everyone thinks the Eagles are a solid top 10 team. Philadelphia is 7th in the preseason rankings.
However, the Eagles have issues to worry about. QB Nick Foles can’t possibly throw interceptions at a lower rate than he did last season. Moreover, the pass defense finished 23rd last season in my yards per pass attempt adjusted for schedule stat.
How well do the predictions compare with the line?
We can check how closely the predictions from these rankings compare with the line. Here are the predictions for week 1. The games are ranked by the strength of the two teams and expected closeness of the outcome.
1. Green Bay at Seattle. (0.61)
Seattle (1) will beat Green Bay (6) by 6.1 at home. Green Bay has a 34% chance of beating Seattle.
2. Indianapolis at Denver. (0.53)
Denver (2) will beat Indianapolis (8) by 7.1 at home. Indianapolis has a 31% chance of beating Denver.
3. San Diego at Arizona. (0.52)
Arizona (12) will beat San Diego (11) by 1.9 at home. San Diego has a 45% chance of beating Arizona.
4. Cincinnati at Baltimore. (0.50)
Baltimore (13) will beat Cincinnati (9) by 0.5 at home. Cincinnati has a 49% chance of beating Baltimore.
5. Carolina at Tampa Bay. (0.43)
Tampa Bay (20) will beat Carolina (15) by 0.6 at home. Carolina has a 48% chance of beating Tampa Bay.
6. New York Giants at Detroit. (0.40)
Detroit (17) will beat New York Giants (23) by 5.0 at home. New York Giants has a 36% chance of beating Detroit.
7. New Orleans at Atlanta. (0.38)
New Orleans (3) will beat Atlanta (18) by 5.0 on the road. Atlanta has a 36% chance of beating New Orleans.
8. Minnesota at St. Louis. (0.36)
St. Louis (19) will beat Minnesota (26) by 5.0 at home. Minnesota has a 36% chance of beating St. Louis.
9. New England at Miami. (0.34)
New England (4) will beat Miami (21) by 5.9 on the road. Miami has a 34% chance of beating New England.
10. Washington at Houston. (0.32)
Houston (28) will beat Washington (25) by 0.8 at home. Washington has a 48% chance of beating Houston.
11. Tennessee at Kansas City. (0.31)
Kansas City (16) will beat Tennessee (27) by 7.5 at home. Tennessee has a 30% chance of beating Kansas City.
12. San Francisco at Dallas. (0.29)
San Francisco (5) will beat Dallas (24) by 7.0 on the road. Dallas has a 31% chance of beating San Francisco.
13. Cleveland at Pittsburgh. (0.23)
Pittsburgh (14) will beat Cleveland (30) by 9.8 at home. Cleveland has a 25% chance of beating Pittsburgh.
14. Jacksonville at Philadelphia. (0.22)
Philadelphia (7) will beat Jacksonville (29) by 12.3 at home. Jacksonville has a 20% chance of beating Philadelphia.
15. Buffalo at Chicago. (0.20)
Chicago (10) will beat Buffalo (31) by 11.6 at home. Buffalo has a 22% chance of beating Chicago.
16. Oakland at New York Jets. (0.18)
New York Jets (22) will beat Oakland (32) by 8.9 at home. Oakland has a 27% chance of beating New York Jets.
The predictions differ the most from the line in games with really bad teams. For example, the preseason rankings predict a 10 point win for Pittsburgh over 30th ranked Cleveland. The line only favors Pittsburgh by 6.5.
There are similar differences for games with Oakland, Buffalo and Jacksonville. It seems like the markets do not want to put down the worst teams in the NFL.
Members of The Power Rank have access to these predictions for all 256 games of the NFL season. To learn more, click here.
Maurice VanGarris says
Way to many homes teams winning their opening day game home vs away average win total is 10/6 , 9/7 last ten years best was 12/4 during that span.
Ed Feng says
Home teams win about 56% of games on average. Unless you think more top teams get home games week 1 by a scheduling quirk, hard to think that home teams will win more than that.
Plus, all the worst teams in the league (Cleveland, Oakland, Buffalo, Jacksonville) are on the road week 1.
What is the figure in the parenthesis referring to ? e.g. in the Green Bay at Seattle. there is a (0.61) listed.
Also at the start of the article where it mentions NFL teams performance regress to the mean…..(note:I’m still trying to get my ahead around regressing to the mean) ….so in terms of the 2 variables regressing to the mean , are you looking at last seasons results as variable 1 and the coming season as variable 2?
Ed Feng says
The number in () is a game rating. Games are ranked on the strength of teams and closeness of expected outcome.
You’re right about regression. If I rate a team 10.0 points better than NFL average one season, expect that rating to decline next season.