How Turnover Margin Affects Win Probability and Win Totals in the NFL

Multiple studies have examined the strong correlation between turnover margin and win probability. We break down the results and tell you what they mean for 2021 NFL betting.

Shawn Wronka
Posted: Sep 8, 2021 1:30 PM ET Updated: Sep 9, 2021 8:27 AM ET Est Time: 4 min
Taylor Stallworth NFL Dallas Cowboys
Photo By - USA TODAY Sports

Every season, bettors everywhere make the assumption that NFL teams that were good and bad the previous season (barring huge roster turnover) are largely going to perform similarly in the upcoming season. However, as some more well-versed bettors know, there are some key indicators that can help identify which teams over- and underperformed in the win column. One of those key indicators is turnover margin.

Turnover margin at a glance

• Teams that win the turnover margin wins the game almost 70% of the time

• Since 2016, teams that had a double-digit positive turnover margin saw an average decrease of 10.5 in their margin in the following year

• Turnovers are predictive of future performance

The importance of turnovers

In a study conducted by the Harvard Sports Analysis Collective, it was found that a team that wins the turnover margin in a game wins 69.6 percent of the time. To demonstrate the relative strength of turnover influence, home teams won “just” 57.2 percent of all games in the same sample. The effect is compounding as well, as teams that won the turnover margin by two or more won 83.9 percent of the time, and 90.7 percent of the time when winning the margin by three or more.

In addition to that, the study found the effect to be rather significant on a seasonal basis. Each additional positive turnover towards a team's margin is worth about 0.2 wins with an R-squared value of 0.419. For the less statistically-inclined, that means that while ignoring all other factors that impact winning and losing (point margin, opponents faced, etc.), turnover margin explained 41.9 percent of the variation in regular-season wins.

Applying this concept to the 2020 season, the top-12 teams that led in turnover margin had winning records, and none had less than 10 wins. Conversely, none of the bottom ten teams in turnover margin had a winning record.

Team-to-team effect

Zooming a little farther out, teams since 2016 that had a double-digit positive turnover margin saw an average decrease of 10.5 in their margin in the following year. Those same teams went from averaging 11.2 wins in their turnover-friendly seasons to averaging 9.9 wins in the following season, a 1.3 decrease in wins. Only one of the 22 teams to post a +10 or higher turnover margin during this stretch did it again the following year.

As you can see below, the 2020 turnover overperformers to watch for are the Titans and Colts.

Team Year To MGN To MGN in Y+1 DIF Wins Wins in Y+1 DIF
KC 2016 16 15 -1 12 10 -2
OAK 2016 16 -14 -30 12 + -6
NE 2016 12 6 -6 14 13 -1
ATL 2016 11 -2 -13 11 10 -1
MIN 2016 11 5 -6 8 13 5
BAL 2017 17 -3 -20 9 10 1
KC 2017 15 9 -6 10 12 2
LAX 2017 12 1 -11 9 12 3
PHI 2017 11 -6 -17 13 9 -4
DET 2017 10 -5 -15 9 6 -3
JAX 2017 10 -12 -22 10 5 -5
SEA 2018 15 12 -3 10 10 0
HOU 2018 13 0 -13 11 10 -1
CHI 2018 12 0 -12 12 8 -4
LAR 2018 11 0 -11 13 9 -4
NE 2018 10 21 11 11 12 1
NE 2019 21 3 -18 12 7 -5
NO 2019 15 9 -6 13 12 0
GB 2019 12 7 -5 13 13 1
SEA 2019 12 4 -8 11 12 -3
MIN 2019 11 -1 -12 10 7 -3
BAL 2019 10 4 -6 14 11
TEN 2020 11 11
IN 2020 10 11
AVG 12.86 2.41 -10.45 11.23 9.86 -1.36

Conversely, since 2016, teams that had a double-digit negative turnover margin saw an average increase of 15.8 (!!!) turnovers in their margin in the following year. Those teams went from averaging 4.4 wins to 7.1 in the following season, a 2.7 (!!!) increase in wins. Only three of the 22 teams to post a -10 or lower turnover margin during this stretch did it again the following year.

The teams to keep an eye on this year are the Eagles, Raiders, 49ers, and Broncos.

Team Year To MGN To MGN in Y+1 DIF Wins Wins in Y+1 DIF
LAR 2016 -11 7 18 4 11 7
CLE 2016 -12 -28 -16 1 0 -1
JAX 2016 -16 10 26 3 10 7
CHI 2016 -20 0 20 3 5 2
NYJ 2016 -20 -10 10 5 5 0
HOU 2017 -12 13 25 4 11 7
MIA 2017 -14 5 19 6 7 1
OAK 2017 -14 -7 7 6 4 -2
DEN 2017 -17 7 24 5 6 1
CLE 2017 -28 7 35 0 7 7
NYJ 2018 -10 -4 6 4 7 3
PIT 2018 -11 8 19 9 8 -1
ARI 2018 -12 -1 11 3 5 2
JAX 2018 -12 -1 11 5 6 1
TB 2018 -18 -13 5 5 7 2
SF 2018 -25 4 29 4 13 9
MIA 2019 -10 9 19 5 10 5
TB 2019 -13 8 21 7 11 4
CAR 2019 -14 1 15 5 5 0
CIN 2019 -14 -7 7 2 4 2
LAC 2019 -17 3 20 5 7 2
NYG 2019 -17 0 17 6 6 0
PHI 2020 -10 4
LV 2020 -11 8
SF 2020 -11 6
DEN 2020 -16 5
AVG 15.32 0.50 15.82 4.41 7.05 2.64

Mid-season swings

But the effect isn’t just at season-end, it can also be a signal of regression in-season as well. In a separate study done by M. Beuoy, it was found that teams with strong season-to-date records with large turnover margins are likely to regress to the mean (and the inverse being true about weak teams with weak turnover margins).

For example, teams that had a 5+ turnover margin relative to their opponents (for example, the Saints with a +3 turnover margin playing the Falcons with a -2 margin) covered 52.0 percent of the time. And again, the effect was compounding: when it was 10+ they covered 54.4 percent of the time and when it was 15+ they covered 55.8 percent of the time.

Are turnovers predictive?

So, we know over- and underperforming in the turnover category is predictive of future performance, but are turnovers themselves predictable?

As an extension of Harvard Sports Analysis Collective's findings, they modeled the likelihood of a turnover being attributed to talent as opposed to luck borrowing some methodology from Tom Tango, a sabermetrician. The findings showed that about 46 percent of all turnovers could be attributed to talent (quarterbacks with low/high INT percentages, defensive linemen particularly proficient at strip-sacks, ballhawk defensive backs, etc.) and the remaining 54 percent to luck (tipped interceptions, the way the ball bounces on a fumble, etc.).

Thus, in summation, fumbles themselves are not very predictable. Given that we know that each incremental fumble towards a team's fumble margin is equal to about 0.2 wins, this creates both positive and negative performances that are extremely hard to repeat at both ends of the turnover margin spectrum. This contributes to misled optimism and pessimism towards these extreme performers when it comes to predicting the future success and failure of teams.

Popular Content

Covers 25 Years Logo Established in 1995,
Covers is the world
leader in sports
betting information.
Covers is verified safe by: Evalon Logo GPWA Logo GDPR Logo GeoTrust Logo Evalon Logo