Flying Under the Radar

Against the Pack
Against the Pack

While the team names might be big, there is something about them that is not detected. Maybe it has to do with their win totals?

The NFL season starts today, so I decided to check the SSA NFL Win Total Model to see if there were any quality win total plays still out there. I noticed something that was a little concerning; the model was most confident in picking against quality teams. Going against the Packers, Seahawks, and Patriots certainly makes me nervous and had me questioning our own work. To better understand why the model was picking against these teams, I decided to conduct a little research and was very surprised at my findings.

The Packers, Seahawks, and Patriots all have win totals over 10. Additionally, all win total lines are offering plus money on the under. The NFL is certainly the most difficult sport to predict with only 16 games, parody in the league, and to quote Bob Voulgaris, “even the ball isn’t round.” So I began to realize the logic of the model. When the future is uncertain, gravitate towards the middle, or in the case of the NFL, towards 8 wins.

While I understood the logic of the model, I wanted to see how the model worked in action. Fortunately at SSA, we have NFL win total lines going back to 2008. I discovered that in the past eight season there have been 21 teams with win total lines of 10 or more that offered plus money on the under. How did they fare? Well, 14 teams went under their win total while 7 went over. Factor in that you are getting plus money on these plays and the observed rate of return comes out to 48 percent!

Now a sharp analytic eye will point out that 21 observations is a limited sample size. So I ran a test to see how likely it would be to observe 14 unders if the true likelihood of occurrence was 50 percent. The probability of this occurring (p-value) is 9.5 percent, indicating that such observations aren’t likely. Additionally, if you account for the juice, the implied win prob for these plays is not .5, but rather .45 (because of the plus money on the under). Re-running the calculations, we arrive at a p-value of 3.8 percent, below the standard 5 percent threshold that is often used in statistical tests.

With the reassurance of my research combine with the SSA Win Total Model, I have three plays to share with you if you’re still looking for some NFL win total action.

Packers Under 11 Wins (+115). Getting +115 and you can only lose if the Packers win 12 or more games? Combine that with the fact that they are the #1 public team in the NFL and the SSA model says there is a lot of value. 3 for 3 on the checklist. Don’t worry Aaron, I double checked.

Seahawks Under 10.5 Wins (+145). Yeah, the Seahawks could win 11 games, but you’re getting +145 on the under. Pete, that’s my deal, what’s yours?

Patriots Under 10.5 Wins (+125). Going against Belichick is not something I’d like to do, but at 10.5 wins and plus money on the under, there is value. Not to mention taking plus money unders on the Pats has proven profitable in recent history. Instead of deflating balls, maybe the Pats should deflate their win total line.

About Jordan Sundheim 185 Articles
Jordan Sundheim is a veteran of the sports analytics field with a background in both athletics and statistics. He earned his master's at Duke, coached football at Stanford, and is currently a sports analytics consultant.