NFL

NFL Betting: Can We Rely on Prior Season Efficiency to Identify Team Win Total Bets?

How much can we rely on last season's data to pick our win totals for the upcoming NFL season?

I certainly have a habit of complicating things.

I like go the extra mile, to research heavily before an answer, to build out more and more detailed betting models before making any bets or recommendations, but sometimes that's not always necessary.

And, to be 100% honest, this ended up being one of those times.

I already had written the introduction to this piece about not complicating things and then went ahead and complicated the whole damn thing anyway.

But I wanted to apply a more simplified approach to NFL team win total over/unders than I would otherwise.

Are there any obvious team-level trends that we can use to anticipate which teams will go over or under their preseason win totals for 2021?

Here's what I found when digging back into win totals since 2012 (using data from Pro-Football-Reference).

Buckets

For this specific look, I wanted to bucket teams into tiers of efficiency so that it's easier to describe what I found.

What did I find, you ask? Well, what I found is that we can totally learn a lot from last year but that -- unsurprisingly -- history doesn't just repeat itself in the high-variance NFL.

At numberFire, we have a metric called nERD, which indicates expected point differential on a neutral field against an average opponent. So, a team with a nERD of 3.00 would be favored by 4.00 points against a team with a nERD of -1.00 at a neutral venue.

We also have a metric called Net Expected Points (NEP), which helps uncover underlying successes and failures of a team.

To help understand it a little better, consider that the NFL average yards per play rate in 2020 was 5.6. Now, imagine a three-yard gain.

That's subpar, yeah? And what about a 10-yard gain? That's a better-than-average play, right?

Well, when factoring in context, it may not be true. A three-yard gain on 3rd-and-2 is a better play than a 10-yard gain on 3rd-and-20. Over a thousand plays, give or take, on the season, this all adds up.

We can then measure each team's offensive and defensive success not by points scored or allowed or yards per play but rather this underlying measure of expectation.

This is key because I effectively want to ask: do good offenses with bad defenses tend to overperform the following year? Things like that, you know?

Here's what I found out.

By the way, I'm removing all teams that pushed on their win total from the conversation.

Let's Talk About Over/Unders

Here are hit rates on overs by bucketed team success level (offensively and defensively) during the actual season when they ranked as such.

(Keep in mind that historical over rates are not 50.0% but actually 47.1%. This can be explained by the fact that preseason win totals don't always even out to 8.0 wins during these 16-game seasons. The average preseason over/under in this sample is actually 8.2 wins.)

In-Season Win Over%
By Unit Bucket
Offense Defense
Top 10 71.9% 67.3%
Average 40.6% 41.9%
Bottom 10 19.8% 22.4%


Okay, so teams that wind up with top-10 in-season offenses hit their over 71.9% of the time, similar to the rate for top-10 defensive units (67.3%). This just means that high-performing teams win games. That checks out.

Meanwhile, the bottom-tier offenses and defenses, unsurprisingly, fail to hit the over even 25.0% of the time.

This isn't shocking news. Teams with strong outputs win games, and teams with poor offense and/or defense typically don't.

But what do these teams by unit strength do the following year? That's what we're here for, after all.

We won't know exactly how teams will rank in 2021, but can we use their 2020 ranks to examine out over odds? Has history shown that ability?

Sort of. In a way.

Next Season Win Over%
By Unit Bucket
Offense Defense
Top 10 46.3% 45.9%
Average 47.2% 45.5%
Bottom 10 49.3% 52.1%


Besides the fact that unders are just better bets long-term (assuming the juice is fair), we can see pretty evident differentials if we just trust that horrible defenses won't be quite as horrible the following season. Bottom-tier defensive squads alone hit the over more than half the time the following year.

But there's a lot more to it.

Here's a matrix-style breakdown. (Again, this removes all teams that pushed on their following season win total).

Next Season
Over Hit %
(Minus Pushes)
Top-10 Defense Average Defense Bottom-10 Defense
Top-10 Offense 39.3% 48.6% 52.4%
Average Offense 53.3% 39.3% 48.4%
Bottom-10 Offense 44.4% 48.1% 57.1%


An example of how to read this properly: teams that had a top-10 offense and a top-10 defense hit their over just 39.3% of the time the following season.

This, intuitively, would be because they overachieved in one year and are hyped up the following season, failing to meet high win total expectations.

Another example: teams that were bottom-10 in both offense and defense hit their over 57.1% of the time the following season. This could be credited to horrible results in the season prior, low expectations the following season, and natural regression toward the NFL average -- with possibly legitimate, tangible personnel improvements.

I could keep on going and going with more granular data, but we're not going to solidify our bets with tiny samples of teams that were top-8 in passing offense but bottom-third in rushing offense and yet top-10 in total defense or anything like that.

And even if the stars align for certain teams and we have concerns about them, we're not going to bet them just because they fit a certain bucket.

This whole operation was your idea meant mostly to see if anything clear jumped off the page, and I think it does.

Here's how those aforementioned buckets compare if we measure them relative to the average over hit rate (again, 47.1%). Think of this as "how much more likely than average have these team buckets been to hit their over?"

Next Season
Over Hit %
Compared to Average
(Minus Pushes)
Top-10 Defense Average Defense Bottom-10 Defense
Top-10 Offense -7.8% 1.4% 5.3%
Average Offense 6.2% -7.8% 1.3%
Bottom-10 Offense -2.7% 1.0% 10.0%


Again, it's largely the underperforming defenses that outperform the overall average while the totally average teams from one year underperform the following season.

Which Teams Fit Where Entering 2021?

With all that said, here is how each squad ranked during the 2020 season by our Adjusted NEP metrics.

We can, perhaps, use this as a guide, but always be mindful of offseason changes that will have impacted teams substantially.

Next Season
Over Hit %
Compared to Average
(Minus Pushes)
Top-10 Defense Average Defense Bottom-10 Defense
Top-10 Offense BAL, NO, TB ARI, BUF, GB,
KC, SEA
CLE, LV,
MIN, TEN
Average Offense IND, LAR, MIA,
PIT, SF
ATL, CAR,
LAC, NE
DAL, DET, HOU
Bottom-10 Offense CHI, WSH DEN, NYG, PHI CIN, JAC, NYJ


History is imploring us to be wary of win totals for the Baltimore Ravens (win total of 11.0 with -120 odds on the under, per FanDuel Sportsbook), the New Orleans Saints (under 9.0 [-105]), and the Tampa Bay Buccaneers (under 11.5 [+125]). numberFire's projections have the Ravens winning 10.9 games, the Saints winning 9.7, and the Buccaneers winning 10.8.

As for the middle of the matrix, we have the Atlanta Falcons (under 7.5 [+110]), Carolina Panthers (under 7.5 [-130]), Los Angeles Chargers (under 9.5 [-135]), and New England Patriots (under 9.0 [+115]). numberFire expects them to win 7.7, 7.8, 9.0, and 8.1 games, respectively.

Teams fitting the heavy over bucket include the Cincinnati Bengals (over 6.5 [+120]), Jacksonville Jaguars (over 6.5 [-110]), and New York Jets (over 6.0 [-115]). numberFire's initial sims have the Bengals winning 7.2 games, the Jaguars winning 6.1, and the Jets winning 6.2.

Perhaps the most appealing tier of all would be the Cleveland Browns (over 10.5 [-105]), Las Vegas Raiders (over 6.5 [-145]), Minnesota Vikings (over 8.5 [-170]), and Tennessee Titans (over 9.0 [-150]), but they're either getting heavy juice or -- in Cleveland's case -- a double-digit winner.