NFL predictions - Adding turnover margin to Pythagorean Theorem to predict regular-season win totals
Welcome to the third part of our series on predicting regular-season win totals in the NFL. In this article, we will look at how we can build on the second part of the series using the Pythagorean Theorem to evaluate a team’s performance and expected win percentage. We will now refine those projections by factoring in turnover margin, and the points scored off turnovers.
If you haven’t read the first part of the series using a team’s record in one-score games to predict regular-season win totals, I suggest you take a look at that before continuing.
There are so few possessions in an NFL game that turnovers matter. Looking at data from Pro Football Reference, if a team has even one more turnover than the other team, they will win 69.6%.
Across a season, each positive turnover is worth about 0.2 wins, and over 40% of a regular-season win variance can be down to turnovers.
So, all you need to do is turn the ball over a lot each season to win more games, right? Well, it’s not that easy. Research shows that around 50% of turnovers are down to pure luck – think the way the ball bounces, receivers falling, tipped passes, etc.
What this means from a betting perspective is that teams who had bad luck last year will see some regression to the mean and vice versa for teams who had good luck. Let’s take a look at some of the last years, top performers, in both categories:
- New England Patriots +21
- New Orleans Saints +15
- Green Bay Packers +12
- Seattle Seahawks +12
- Minnesota Vikings +11
- Baltimore Ravens +10
- Los Angeles Chargers -17
- New York Giants -17
- Cincinnati Bengals -14
- Carolina Panthers -14
- Tampa Bay Buccaneers -13
- Miami Dolphins -10
Now let’s get into using turnover margin with the Pythagorean Theorem to strengthen our season win projections. To do this, you multiply the team’s turnover margin by 3.2 to give us the number of points generated by turnovers. You then subtract this figure from their points scored and plug the new points scored number into the Pythagorean Theorem calculation.
Here are the top underperformers from 2019, which includes their record and what their projected number of wins should have been in parentheses.
- LA Chargers 5-11 (9 wins)
- Cincinnati Bengals 2-14 (5.96 wins)
- Detroit Lions 3-12-1 (6.66 wins)
- New York Giants 4-12 (6.95 wins)
- Dallas Cowboys 8-8 (10.40 wins)
The Chargers topped the list this time around with a whopping 4 games under expectation. Their 2020 win total opened 7.5 and has been bet up to 8 in places, but if you shop around, there’s still some 7.5’s about.
Now let’s take a look at the teams who were lucky last season:
- Green Bay Packers 13-3 (8.6 wins)
- Seattle Seahawks 11-5 (7.34 wins)
- New Orleans Saints 13-3 (9.46 wins)
- Houston Texans 10-6 (7.85 wins)
- San Francisco 49ers 13-3 (11.09 wins)
- Baltimore Ravens 14-2 (12.13 wins)
Green Bay was the luckiest team last year with a 4.4 game difference in expected wins. Their 2020 win total varies between 8.5 and 9 depending on the book, but it looks like the regression has already been priced in by oddsmakers.
Full Pythagorean Theorem with turnover margin standings
Here’s how the full NFL standings should have looked in 2019 based on the Pythagorean Theorem with turnover margin added (teams actual record in parenthesis).
Chiefs 10.50 wins (12-4)
Chargers 9.00 wins (5-11)
Broncos 7.00 wins (7-9)
Raiders 5.88 wins (7-9)
Patriots 11.37 wins (12-4)
Bills 9.20 wins (10-6)
Jets 6.29 wins (7-9)
Dolphins 5.10 wins (5-11)
Ravens 12.13 wins (14-2)
Browns 7.31 wins (6-10)
Steelers 6.89 wins (8-8)
Bengals 5.96 wins (2-14)
Titans 9.02 wins (9-7)
Texans 7.85 wins (10-6)
Colts 7.6 wins (7-9)
Jaguars 5.89 wins (6-10)
49ers 11.09 wins (13-3)
Rams 8.63 wins (9-7)
Seahawks 7.34 wins (11-5)
Cardinals 6.47 wins (5-10-1)
Cowboys 10.40 wins (8-8)
Eagles 8.87 wins (9-7)
Giants 6.95 wins (4-12)
Redskins 4.28 wins (3-13)
Vikings 9.61 wins (10-6)
Packers 8.60 wins (13-3)
Bears 7.50 wins (8-8)
Lions 6.66 wins (3-12-1)
Saints 9.46 wins (13-3)
Buccaneers 8.85 wins (7-9)
Falcons 7.96 wins (7-9)
Panthers 6.42 wins (5-11)
From this analysis, some teams stand out to me as regression candidates, and there still looks like there could be some betting value. These will be added to the shortlist, but I won’t bet them until we dig into the numbers a bit further, analyze the 2020 schedules, and look at personnel changes, but for reference they are:
Seahawks Under 9.5
Chargers Over 7.5
Raiders Under 7.5
Next up, we’ll look at teams strength of schedule and personnel changes in detail to see if we can isolate some value.
Bookmark our NFL expert picks homepage for free picks on every game of the 2020 NFL season.