«

»

Aug 20

Updating the Expected Win Differential for 2015

NFL Preview

Once again, our Expected Win Differential formula effectively predicted improvement or decline for teams that finished outside of the threshold of regression. That makes it important to determine EWD results for 2015 to help determine which teams are likely to take step forward or step back this upcoming season. Hence, the NFL Preview features section will come to a close with a look at EWD.

Before the 2015 results are revealed, let’s quickly trace back to what went right and wrong in 2014. The results were as followed:

  • Houston Texans (+3.78 EWD): Improved from 2-14 to 9-7
  • Indianapolis Colts (-2.62 EWD): Owned 11-5 record both seasons
  • Detroit Lions (+2.45 EWD): Improved from 7-9 to 11-5
  • Atlanta Falcons (+2.44 EWD): Improved from 4-12 to 6-10
  • Washington Redskins (+2.41 EWD): Improved from 3-13 to 4-12
  • New England Patriots (-2.21 EWD): Owned 12-4 record both seasons
  • Cleveland Browns (+2.09 EWD): Improved from 4-12 to 7-9
  • Seattle Seahawks (-1.76 EWD): Declined from 13-3 to 12-4
  • Philadelphia Eagles (-1.59 EWD): Owned 10-6 record both seasons
  • Minnesota Vikings (+1.55 EWD): Improved from 5-10-1 to 7-9
  • New York Jets (-1.52 EWD): Declined from 8-8 to 4-12

Eight of the 11 teams regressed to some degree in win-loss record as expected. The other three teams maintained their record as is. This gives EWD a nearly identical success rate to last year. So far, through 22 teams over a two-year span, 15 regressed and seven maintain their record. None defied the expected regression with either: (1) an improved record after a negative-1.454 EWD or worse, or (2) a declined record after a 1.454 EWD or better. Meanwhile, the correlation between each team’s 2013 true wins and 2014 true wins was 0.5926, while the correlation between each team’s 2013 EWD-adjusted wins and 2014 true wins was 0.6649. Despite little room for EWD to outshine an unusually stagnant year, it still did. This happened even with the four division winners above declining only a combined one game. Chalk it up as a “2-for-2” start for the Expected Win Differential formula.

Now let’s focus on 2015. To calculate the Expected Win Differential and determine which teams may be in line for win-loss record regression, we must take three steps. The first step involves calculating the Pythagorean Win Differential (PWD) of each team (Table 1). Remember that the Pythagorean win formula used here is slightly different that what was originally constructed, as a result of the league’s uptick in scoring. The results of this step can give you an idea which teams are most at risk for regression.

Table 1: Pythagorean Win Differential

Team PythW Record Differential Team PythW Record Differential
BAL 10.79 10-6 +0.79 wins CHI 5.02 5-11 +0.02 wins
CIN 8.57 10-5-1 -1.93 wins DET 9.23 11-5 -1.77 wins
CLE 6.86 7-9 -0.14 wins GB 11.05 12-4 -0.95 wins
PIT 9.61 11-5 -1.39 wins MIN 7.48 7-9 +0.48 wins
BUF 9.62 9-7 +0.62 wins DAL 10.61 12-4 -1.39 wins
MIA 8.38 8-8 +0.38 wins NYG 7.51 6-10 +1.51 wins
NE 11.59 12-4 -0.41 wins PHL 9.61 10-6 -0.39 wins
NYJ 4.84 4-12 +0.84 wins WSH 4.62 4-12 +0.62 wins
HOU 9.81 9-7 +0.81 wins ATL 7.14 6-10 +1.14 wins
IND 10.03 11-5 -0.97 wins CAR 7.06 7-8-1 -0.44 wins
JAX 3.68 3-13 +0.68 wins NO 7.47 7-9 +0.47 wins
TEN 3.41 2-14 +1.41 wins TB 4.49 2-14 +2.49 wins
DEN 10.83 12-4 -1.17 wins ARZ 8.35 11-5 -2.65 wins
KC 10.14 9-7 +1.14 wins SEA 11.86 12-4 -0.14 wins
OAK 3.18 3-13 +0.18 wins SF 6.99 8-8 -1.01 wins
SD 8 9-7 -1 win STL 7.15 6-10 +1.15 wins

PythW: Pythagorean Wins, or (Points Scored ^ 2.4) / ((Points Scored ^ 2.4) + (Points Allowed ^ 2.4))

At first glance, it seems like the NFC had a more zany year than the AFC. Many publications simply use Pythagorean Win Differential as the means to predict win-loss regression, and they do so with teams that pass the one-win threshold. Using those parameters, there would be six AFC teams and eight NFC teams due for regression. Furthermore, both teams that doubled the threshold are in the NFC, making the possibility of some big changes in the conference in 2015.

Of course, there are some other factors of context to account for beyond PWD. We like to use Turnover Win Impact regression (TWIr) as a contextual factor, so we need to examine that as well (Table 2). This is the second step in today’s study. Generally speaking, turnover margin is hard to maintain from one year to the next. There are obvious exceptions to that rule, which likely involves a team with an elite quarterback who protects the ball superbly well. As Jim Armstrong pointed out a decade ago for Football Outsiders, offensive turnover rates correlate better from year-to-year than defensive turnover rates. Keep that in mind when reading the remainder of this study.

Looking at 2014, turnover margin had a surprisingly low impact. No team averaged a net turnover per game in either direction last year. Green Bay owned the best turnover margin (+14, or +0.875 per game), while Oakland owned the worst (-15, or -0.9375 per game). This is extremely rare, with 1995 being the only other 16-game season in which this happened. In fact, most 16-game seasons involve at least one team with a plus-16 margin or better and at least one team with a minus-16 margin or better. Only 1993, 1994, 1995 and 2014 saw no teams surpass at least a plus-16 margin, while only 1980, 1995 and 2014 saw no teams surpass at best a minus-16 margin.

Note that we made one adjustment to the TWIr formula, as 2014-scoring-based PWD results dictate a 38 points-per game-differential equivalency. That means the formula reverted back to what was used two years ago.

Table 2: Turnover Win Impact Regression

Team Margin Points TWIr Team Margin Points TWIr
BAL +2 +8 +0.17 wins CHI -5 -20 -0.43 wins
CIN Even 0 +0 wins DET +7 +28 +0.60 wins
CLE +6 +24 +0.52 wins GB +14 +56 +1.20 wins
PIT Even 0 +0 wins MIN -1 -4 -0.09 wins
BUF +7 +28 +0.60 wins DAL +6 +24 +0.52 wins
MIA +2 +8 +0.17 wins NYG -2 -8 -0.17 wins
NE +12 +48 +1.03 wins PHL -8 -32 -0.69 wins
NYJ -11 -44 -0.95 wins WSH -12 -48 -1.03 wins
HOU +12 +48 +1.03 wins ATL +5 +20 +0.43 wins
IND -5 -20 -0.43 wins CAR +3 +12 +0.26 wins
JAX -6 -24 -0.52 wins NO -13 -52 -1.12 wins
TEN -10 -40 -0.86 wins TB -8 -32 -0.69 wins
DEN +5 +20 +0.43 wins ARZ +8 +32 +0.69 wins
KC -3 -12 -0.26 wins SEA +10 +40 +0.86 wins
OAK -15 -60 -1.29 wins SF +7 +28 +0.60 wins
SD -5 -20 -0.43 wins STL -2 -8 -0.17 wins

TWIr: 0.8175 * (“Points” / 38)

Several teams with notable turnover margins don’t surprise us. Green Bay (plus-14) and New England (plus-12) have the quarterbacks to dominate and take care of the football. Oakland (minus-15) is simply continuing a multi-year stretch of undisciplined football. We’d say the biggest candidates for turnover regression are Houston (plus-12) and New Orleans (minus-13).

The final step in today’s study involves putting it all together. We need to calculate the Expected Win Differential (EWD) for each team in 2015 (Table 3). If the team surpasses the regression threshold of ±1.454 wins, we know to take a deeper examination into their prospects of improvement or decline. Teams in line for improvement are denoted in blue, while team in line for decline are denoted in red.

Table 3: Expected Win Differential

Team PWD TWIr EWD Team PWD TWIr EWD
BAL +0.79 +0.17 +0.62 wins CHI +0.02 -0.43 +0.45 wins
CIN -1.93 0 -1.93 wins DET -1.77 +0.60 -2.37 wins
CLE -0.14 +0.52 -0.66 wins GB -0.95 +1.20 -2.15 wins
PIT -1.39 +0 -1.39 wins MIN +0.48 -0.09 +0.57 wins
BUF +0.62 +0.60 +0.02 wins DAL -1.39 +0.52 -1.91 wins
MIA +0.38 +0.17 +0.21 wins NYG +1.51 -0.17 +1.68 wins
NE -0.41 +1.03 -1.44 wins PHL -0.39 -0.69 +0.30 wins
NYJ +0.84 -0.95 +1.79 wins WSH +0.62 -1.03 +1.65 wins
HOU +0.81 +1.03 -0.22 wins ATL +1.14 +0.43 +0.71 wins
IND -0.97 -0.43 -0.54 wins CAR -0.44 +0.26 -0.70 wins
JAX +0.68 -0.52 +1.20 wins NO +0.47 -1.12 +1.59 wins
TEN +1.41 -0.86 +2.27 wins TB +2.49 -0.69 +3.18 wins
DEN -1.17 +0.43 -1.60 wins ARZ -2.65 +0.69 -3.34 wins
KC +1.14 -0.26 +1.40 wins SEA -0.14 +0.86 -1.00 wins
OAK +0.18 -1.29 +1.47 wins SF -1.01 +0.60 -1.61 wins
SD -1 -0.43 -0.57 wins STL +1.15 -0.17 +1.32 wins

EWD: PWD – TWIr; Note: The threshold for regression is ±1.454 wins.

In total, 14 of the 32 NFL teams surpassed the regression threshold in 2014. Nine are in the NFC! This includes Tampa Bay and Arizona, who are the two teams to own a EWD at least twice the value of the regression threshold. We’ll examine each case during our Five Factors of Regression. Know that some cases (Green Bay) are more flimsy than others (Tampa Bay), but we’ll save that discussion for tomorrow and beyond. For now, if your team is in red or blue, just know they’ve been put on notice.

Leave a Reply

%d bloggers like this: