But we should fight the urge toward overconfidence, especially in the face of history. Optimism for these probable cellar dwellers might feel forced. 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. 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. 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. 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. 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. Forecasting a sport as luck-driven as the NFL is rough that way. 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.

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. Last season, the Chicago Bears were initially forecast for seven wins by Las Vegas, then traded for Khalil Mack and somehow won 12. Every year, we look back on preseason win totals produced by forecasters and betting markets and chuckle at some of the more egregious misses.

2019 nfl win totals