Okay, So Not Everyone Is Perfect/NFL Predictions

In my Sunday Modifying Football's Pythagorean Theorem article, I stated that 2.64 was the optimal exponent for football.

Except that it isn't.

Instead of being lazy and only using information since 2002, I gathered every team's record, points scored, and points allowed since the NFL merger in 1970, and tried to find the best exponent for various time periods since then.

Listed along with the exponents, time periods, and method (James' and Smyth's, which were defined in the first article linked above) is the root-mean-square-error of each exponent.

It's calculated by squaring the "error" of each team (in the case, the difference that was used in the previous article), finding the average of the squares, then taking the square root of that number, which I realized was more accurate than using the straight difference.

	James'	RMSE	Smyth's	RMSE
1970-now	2.33 exp	1.462	0.229 exp	1.453
1980-now	2.49 exp.	1.393	0.246 exp.	1.388
1990-now	2.53 exp.	1.225	0.250 exp.	1.215
1970s	1.97 exp.	1.591	0.187 exp	1.582
1980s	2.42 exp.	1.670	0.237 exp.	1.671
1990s	2.50 exp.	1.234	0.248 exp	1.225
2000s	2.45 exp.	1.453	0.240 exp.	1.449

As the table shows, the 1970s and 1980s were the most unpredictable decades, with the 1990s through now as the most predictable—that's good, obviously.

And as always, Smyth's floating exponent was better than the James method in every time period, except the 1980s, which was basically a wash.

So, which exponent should we focus on, the 1990s-now time period, or this decade's, which had a higher RMSE than the former in both methods? Me, I'd stick with a 2.53 and 0.25 exponent for each method, or that of the last two decades.

The following table shows how those two exponents, including the two ideal ones for this decade alone, compared to each NFL team's actual record last year. The "Average Diff." column shows the average difference of each formula and their actual wins; a positive number shows the team in question underperformed, and a negative shows that the team in question overperformed.

Team	Wins	Losses	James '90-now	Smyth '90-now	James 2000	Smyth 2000	Average Diff.
Miami Dolphins	1	15	3.6	3.5	3.7	3.6	2.60
Minnesota Vikings	8	8	9.6	9.6	9.5	9.6	1.58
Pittsburgh Steelers	10	6	11.6	11.6	11.5	11.5	1.52
New York Jets	4	12	5.3	5.3	5.3	5.4	1.33
Philadelphia Eagles	8	8	9.1	9.1	9.1	9.1	1.12
Tampa Bay Buccaneers	9	7	10.1	10.1	10.0	10.0	1.05
Seattle Seahawks	10	6	10.9	10.9	10.8	10.8	0.87
Cincinnati Bengals	7	9	7.9	7.9	7.9	7.9	0.87
Oakland Raiders	4	12	4.7	4.7	4.8	4.8	0.78
New Orleans Saints	7	9	7.8	7.8	7.8	7.8	0.76
Chicago Bears	7	9	7.6	7.6	7.6	7.6	0.59
St. Louis Rams	3	13	3.5	3.4	3.6	3.5	0.48
San Diego Chargers	11	5	11.5	11.6	11.4	11.4	0.48
Kansas City Chiefs	4	12	4.3	4.4	4.4	4.5	0.43
Arizona Cardinals	8	8	8.1	8.1	8.1	8.1	0.13
Jacksonville Jaguars	11	5	10.9	11.0	10.8	10.9	-0.11
Houston Texans	8	8	7.9	7.9	7.9	7.9	-0.13
Baltimore Ravens	5	11	4.8	4.8	4.9	4.9	-0.14
Atlanta Falcons	4	12	3.7	3.7	3.9	3.8	-0.21
Washington Redskins	9	7	8.8	8.7	8.7	8.7	-0.26
Indianapolis Colts	13	3	12.8	12.8	12.6	12.7	-0.27
Green Bay Packers	13	3	11.8	11.8	11.6	11.7	-1.26
New York Giants	10	6	8.6	8.6	8.6	8.6	-1.39
Denver Broncos	7	9	5.6	5.5	5.7	5.6	-1.40
San Francisco 49ers	5	11	3.5	3.6	3.6	3.7	-1.42
Detroit Lions	7	9	5.6	5.4	5.6	5.5	-1.46
Cleveland Browns	10	6	8.5	8.5	8.5	8.5	-1.48
Carolina Panthers	7	9	5.4	5.5	5.5	5.6	-1.50
Dallas Cowboys	13	3	11.2	11.3	11.1	11.2	-1.77
Tennessee Titans	10	6	8.1	8.1	8.1	8.1	-1.87
New England Patriots	16	0	14.0	14.2	13.9	14.1	-1.96
Buffalo Bills	7	9	4.8	4.8	4.8	4.9	-2.17

• Miami underperformed and New England overperformed—but that can be expected with their outlier records. The Bills and Titans, however, aren't good enough to be playoff contenders like they were last year. My 2008 predicted records for them are below.
• Two notes about the Cowboys: One, they can't even win in the playoffs, and two, they even did a lot better than they were expected to do last year. Don't expect them to make the Super Bowl or even the NFC Championship game.
• In fact, I wouldn't be the least surprised to see the Eagles, who should have won a game more than they won last year, beat them out for the NFC East title.
• The Seahawks won almost a game less than they were expected to last year, and as you will see below, I like them a lot this year.

And now, my 2008 NFL Predictions.

AFC East (Team, Predicted wins-losses; * denotes wild card)

New England Patriots, 14-2

New York Jets, 9-7

Buffalo Bills, 5-11

Miami Dolphins, 4-12

AFC South

Indianapolis Colts, 12-4

Jacksonville Jaguars*, 10-6

Houston Texans*, 10-6

Tennessee Titans, 6-10

AFC North

Pittsburgh Steelers, 10-6

Baltimore Ravens, 9-7

Cleveland Browns, 7-9

Cincinnati Bengals, 6-10

AFC West

San Diego Chargers, 11-5

Denver Broncos, 7-9

Oakland Raiders, 5-11

Kansas City Chiefs, 4-12

NFC East

Dallas Cowboys, 10-6

Philadelphia Eagles*, 9-7

Washington Redskins, 9-7

New York Giants, 8-8

NFC South

New Orleans Saints, 10-6

Tampa Bay Buccaneers, 9-7

Carolina Panthers, 6-10

Atlanta Falcons, 6-10

NFC North

Green Bay Packers, 9-7

Minnesota Vikings*, 9-7

Detroit Lions, 6-10

Chicago Bears, 5-11

NFC West:

Seattle Seahawks, 12-4

Arizona Cardinals, 9-7

St. Louis Rams, 7-9

San Francisco 49ers, 5-11

Note that my predictions give out a total of 256 wins and 256 losses, unlike other online predictions. (ESPN's Mike Greenberg predicted a total of 269 wins and 243 losses. Thirteen more wins than will happen!)

I had a hard time choosing the second wild card in the NFC, going back-and-forth between Tampa Bay and Minnesota.

Wild Card Round

Chargers over Houston; Jacksonville over Pittsburgh

Saints over Vikings; Eagles over Packers

Divisional Round

Patriots over Jacksonville; Colts over Chargers (depending on the health of Peyton and Merriman)

Seahawks over Eagles; Saints over Cowboys

Championship Round:

Patriots over Colts/Chargers

Seahawks over Saints

Super Bowl XLIII

New England Patriots over Seattle Seahawks.