When you want to look at the overall strength of an NFL team, most analysts know that there are better metrics than simply each team's win-loss record. The 2017 Bills, for example, won nine games and made the playoffs despite being outscored by 57 points on the season.
When the 2018 win totals market opened up, the Bills weren't being valued like a nine-win team. In fact, they currently sit with the lowest win total of any team (their line is 6, with the Under juiced up to -190, meaning you have to bet $1.90 for every $1 of potential profit).
There are plenty of metrics to dive into to determine how good each team is -- I'm particularly partial to Football Outsiders' DVOA, especially broken down to different parts of a team -- but one easy-to-understand metric, despite its unwieldy name, is Pythagorean wins.
Pythagorean wins takes how many points a team scored and allowed and generated an expected record based on that total. If you give up exactly the same amount of points as you score, meaning you have a point differential of zero, PyWins calls you an 8-8 team. Teams with a positive point differential have more than 8.0 PyWins and vice versa. The 2017 Bills, which we saw earlier had a point differential of minus-57, were a 6.7 PyWin team.
Why is this important? Simply put, PyWins has much more predictive value on the following season than a team's standard record.
Want more picks? At SportsLine, I've posted updated analysis of the Super Bowl betting lines and five win totals I like. Sign up for a subscription with promo code WHITE to get your first month for $1!
Outliers from 2016 deliver 2017 value
For this analysis, I'm going to use Brett Lieblich's Adjusted Pythagorean Theorem, explained over at Football Outsiders. Basically, it's the same as the typical points scored/allowed model, but with garbage time scoring removed, slightly improving its utility.
In that piece, Lieblich shows that five teams in 2016 overperformed their adjusted PyWins number by at least two wins. All had regression built in to their 2017 win totals, but none of them were able to meet those reduced expectations, whether due to injury, ineptitude or a little of both.
| Team | 2016 W | Ad Pyth | Diff | 2017 W | O/U | Result |
|---|---|---|---|---|---|---|
| OAK | 12 | 8.61 | -3.39 | 6 | 10 | Under |
| HOU | 9 | 6.57 | -2.43 | 4 | 8.5 | Under |
| MIA | 10 | 7.73 | -2.27 | 6 | 7.5 | Under |
| NYG | 11 | 8.92 | -2.08 | 3 | 9 | Under |
| DAL | 13 | 10.95 | -2.05 | 9 | 9.5 | Under |
The results for teams that underperformed their adjusted PyWins figures by at least two wins are a mixture of teams that went Over and Under their final win total numbers, but the team at the top is likely the one that took the most people by surprise in 2017.
| Team | 2016 W | Ad Pyth | Diff | 2017 W | O/U | Result |
|---|---|---|---|---|---|---|
| JAC | 3 | 6.03 | 3.03 | 10 | 6.5 | Over |
| SD/LA | 5 | 7.60 | 2.60 | 9 | 7.5 | Over |
| CLE | 1 | 3.55 | 2.55 | 0 | 5 | Under |
| ARI | 7 | 9.46 | 2.46 | 8 | 8.5 | Under |
| CIN | 6 | 8.21 | 2.21 | 7 | 8.5 | Under |
| CHI | 3 | 5.02 | 2.02 | 5 | 5.5 | Under |
| PHI | 7 | 9.00 | 2.00 | 13 | 8 | Over |
While several of those underperforming teams went under their Vegas win total predictions, an overall record of 8-4 is nothing to be ashamed about. And you'll note that every team but the Browns actually saw their record improve or decline in the direction that PyWins suggested, even if they didn't match the Vegas outlook.
Think you're a football expert? It's time to prove it. Compete against friends, family and co-workers or put your football knowledge to the test against the NFL Today cast. Play 2018 Football Pick 'Em now!
What 2017 PyWins data say about 2018
Only four teams saw a difference of two wins or more between their actual win total and adjusted PyWins in 2017.
| Team | 2017 W | Ad Pyth | Diff | 2018 O/U |
|---|---|---|---|---|
| CLE | 0 | 2.92 | 2.92 | 6 |
| BUF | 9 | 6.68 | -2.32 | 6 |
| PIT | 13 | 10.83 | -2.17 | 10.5 |
| JAC | 10 | 12.02 | 2.02 | 9 |
The Browns of course have nowhere to go but up after winning zero games, and the Bills are predictably expected to fall well short of their 2017 win total.
The Steelers are an interesting case, as their 9.26 PyWins from 2016 suggested that an 11-5 record might be expected to fall last year. Instead they won 13 games, making them one of five teams to move against what adjusted PyWins projected, and the only team aside from the Browns to defy a PyWins difference of more than 0.57. Can they do it again this year?
The Jaguars are an even more interesting case, as they're the rare team that the adjustment actually goes against the regression to .500. This could be an indicator that despite owning a win total of 9, they could be in for another season with double-digit wins.
I've actually listed two of these five teams in my favorite win total picks over at SportsLine, including one that would defy the data above, since it's important to look at the context beyond the raw data when evaluating
I mentioned earlier that aside from the Browns and Steelers, most teams moved according to how you'd expect when looking at adjusted PyWins from the previous year, at least once the differential gets big enough. Those two were the only teams out of 21 that had at least a one-win differential to defy that movement. So let's widen our field for 2018 and take a look at the teams who had a PyWins difference between 1.5 and 2 wins in 2017.
| Team | 2017 W | Ad Pyth | Diff | 2018 O/U |
|---|---|---|---|---|
| CAR | 11 | 9.11 | -1.89 | 9 |
| ARI | 8 | 6.25 | -1.75 | 6 |
| HOU | 4 | 5.71 | 1.71 | 8.5 |
| BAL | 9 | 10.64 | 1.64 | 8.5 |
| TEN | 9 | 7.42 | -1.58 | 8 |
| TB | 5 | 6.51 | 1.51 | 6.5 |
Most of these teams have 2018 Vegas win totals that move in ways adjusted PyWins would predict; the question is whether that regression will exceed the Vegas prediction or not. One team like the Jaguars that defies its expected win total trend is the Ravens, who underperformed by 1.64 wins in 2017 yet still managed a 9-7 record. Vegas is expecting them to fall short of nine wins this season, but according to this data, a season with double-digit wins wouldn't be shocking in the least.
