Every year, we look back on preseason win totals produced by forecasters and betting markets and chuckle at some of the more egregious misses. Last season, the Chicago Bears were initially forecast for seven wins by Las Vegas, then traded for Khalil Mack and somehow won 12. The Green Bay Packers’s predicted win total was 10, but they melted down in spectacular fashion and ultimately ended up winning just six games.
We’ve already published our Elo projections, and we think they’re the best we’ve ever produced for the NFL, but there will still be lots of misses to grouse about come January. Forecasting a sport as luck-driven as the NFL is rough that way.
It raises the question: How good are betting markets at predicting team wins? To find out, I got my hands on a tranche of win prediction data stretching back to 1989, courtesy of Sports Odds History, and checked how well Vegas preseason win totals predict actual team wins. While Vegas overall does a good job identifying good and bad teams, it turns out that at the lower end of the range of projected wins, Vegas predictions don’t seem particularly well calibrated — though the confidence intervals at the lower end are large because of the small sample size, so the results aren’t statistically significant.
Projected win totals of six and fewer undersell teams’ prospects by about a win on average, with the exception of teams forecast for five wins.
Win totals don’t change as frequently as the moneyline odds, so we probably shouldn’t take win totals at face value — at least for teams with low projected wins. What does this mean for non-bettors? It should be decent news for the Cincinnati Bengals, New York Giants, Oakland Raiders and Washington Redskins — teams that both Elo and Vegas have pegged for six wins in 2019 — since we should be more bullish on their chances than we currently are.
Optimism for these probable cellar dwellers might feel forced. But we should fight the urge toward overconfidence, especially in the face of history. A few of these teams will end up surprising us — in a good way — at the end of the year for reasons inscrutable to us now.
Which NFL teams might beat expectations?
Average actual wins (1989-2018) by Vegas preseason expected wins, and the 2019 teams at each number of expected wins
|10.5||9.9||Kansas City, L.A. Rams, New Orleans|
|9.5||9.0||Chicago, Green Bay, L.A. Chargers|
|9.0||8.5||Cleveland, Dallas, Indianapolis, Minnesota, Pittsburgh|
|8.5||8.9||Atlanta, Baltimore, Houston, Seattle|
|8.0||7.4||Jacksonville, San Francisco, Tennessee|
|7.0||6.9||Buffalo, Denver, N.Y. Jets|
|6.5||6.4||Detroit, Tampa Bay|
|6.0||6.7||Cincinnati, N.Y. Giants, Oakland, Washington|
Well … maybe not the Bengals. Not only is Cincinnati saddled with an injured A.J. Green, who appears to be out until around Week 8, the Bengals have an offense that is bereft of top talent at nearly every position. Cincinnati replaced head coach Marvin Lewis after 16 seasons of on-again, off-again contention and turned instead to Zac Taylor, a coach best known for being friends with L.A. Rams wunderkind Sean McVay. The hope must be that Taylor can revitalize the career of quarterback Andy Dalton, who sports a middling career yards per attempt of 7.2 and is one of the few starting quarterbacks who Vegas believes wouldn’t move a line if he were to be replaced in the lineup. The defense doesn’t offer a compelling reason for optimism: The Bengals ranked 28th in defensive Defense-adjusted Value Over Average (DVOA) last season. Perhaps we should view that as a reason to be bullish on their prospects in 2019 simply due to regression, since defensive performance year to year isn’t terribly stable. If that seems like a bridge too far, magic might be the answer: Taylor may give lip service to the notion that he isn’t trying to be like his mentor McVay, but McVay’s brand of QB sorcery seems like the best hope for the Bengals to crest seven wins this year.
The Giants are more interesting. After a promising preseason performance by first-round pick Daniel Jones, New York fans are clamoring for a change of the guard at quarterback. As big of a reach as many believed Jones to be, I still see him as a better use of first-round draft capital than “generational talent” Saquon Barkley. Hailed as a potential savior and the missing piece for Eli Manning’s final championship push, Barkley helped the Giants improve from a terrible three-win team in 2017 to a merely bad five-win unit in 2018.
The Giants were second-worst in the league on Expected Points Added per play on first-down play-action passes after adding Saquon to the backfield,1 and prospects for a bounceback in play-action efficiency seem bleak. After trading all-world wideout Odell Beckham Jr. to the Browns, the Giants lost free agent acquisition Golden Tate to a four-game suspension for performance-enhancing drugs and their expected No. 3 wideout Corey Coleman to a season-ending ACL injury. Their best hope for a productive season may rest in ownership’s willingness to bench Manning for good this time.
The other team to somehow accumulate negative value on first-down play action was Oakland. In what seems to be a pattern for teams at the bottom of the win total forecast, Vegas sees Derek Carr as a quarterback worth just 1 point to the spread. The stats back up that view. Carr’s career yards per attempt is, at 6.7, below league average, and his best season as judged by QBR is an anemic 54.6. His weapons are improved from a year ago, but they are volatile. New Raiders wideout Antonio Brown sat out of practice because he wasn’t allowed to wear a helmet the NFL deems dangerous and is now likely to be suspended for some period of time, and Tyrell Williams is a boom or bust weapon who likes to be targeted deep — something Carr may be reluctant to do given his career average depth of target of just 7.7 yards. Meanwhile “Hard Knocks” captured head coach Jon Gruden disparaging “all the football stats and all the fantasy bullshit” in favor of running backs that will “BOOF” the opposing team in pass protection. Of all the six-win teams, Oakland may be the most unpredictable — and that unpredictability could manifest itself in good ways, as well as bad. Brown’s antics could end with a fashionable and safe new helmet, Carr might be coaxed into throwing the deep ball to a talented field stretcher, and Gruden might use rookie running back Josh Jacobs optimally, leading to wins we simply can’t foresee at this point.
The final team projected for six wins in 2019 is Washington, a team that somehow came to the determination that Mark Sanchez and Josh Johnson were better choices than Colin Kaepernick to take over for quarterback Alex Smith when his 2018 season — and perhaps his career — ended with a gruesome leg injury.
In the draft, Washington team president Bruce Allen added Ohio State QB Dwayne Haskins in the first round but then failed to surround him with receiving weapons. Jamison Crowder left via free agency, former first-round bust Josh Doctson was released at the end of the preseason and tight end Jordan Reed suffered another concussion heading into Week 1. Their current starting wide receivers are third-round pick Terry McLaurin — also from Ohio State — and Paul Richardson.
The outlook at running back is brighter with the return of Derrius Guice from an ACL tear that derailed his rookie season, but there is little evidence to suggest they will put him in advantageous spots to run the ball. With the ageless, tackle-breaking cyborg Adrian Peterson in 2018, Washington lined up against neutral or stacked boxes on first-and-10 or second and long 174 times, decided they liked the look and ran right into the scrum 72 percent of the time. But if Washington can flip the script on downs tailor-made for passing and eke out some yards where they should come easy, the duo of Guice and Peterson could be enough to protect current starter Case Keenum or rookie Haskins while he learns on the job — and possibly beat the team’s six-win projection.
Check out our latest NFL predictions.