Week 1 of the 2021 NFL season was deeply strange. Whether it was the Saints thrashing the Packers to start the post-Drew Brees era, the Steelers upending the ballyhooed Bills in Buffalo or the Cardinals stomping the Titans on the road, there were a lot of season-opening results that had us furiously adjusting what we thought we knew about each team in preseason.
As a matter of fact, according to our Elo ratings, it’s hard to find an opening set of games more historically bizarre than what we all just witnessed.
The average Week 1 winner went into its game as an underdog, with just a 49.8 percent chance at the victory — only the fourth time that’s happened in the opening week of a season since the 1970 AFL-NFL merger. And if we flip that around to look at Elo’s favorites, those teams went a mere 6-10 in Week 1, good for a success rate of just 37.5 percent. Since 1970, only one other season’s opening week was tougher on higher-rated squads: 1983, when 10 of 14 favorites fell (including three teams with at least a 69.5 percent chance of winning):
Week 1 of 2021 was historically chaotic
Most unpredictable opening weeks of an NFL season since 1970, based on the lowest average pregame win probability for eventual winners and the lowest winning percentage for pregame favorites (according to Elo)
| Season | Winners’ Avg. Win Prob. | Season | Favorites’ Winning % | |
|---|---|---|---|---|
| 1983 | 44.0% | 1983 | 28.6% | |
| 1980 | 45.9 | 2021 | 37.5 | |
| 1982 | 49.1 | 1980 | 42.9 | |
| 2021 | 49.8 | 1972 | 46.2 | |
| 1978 | 50.2 | 1999 | 46.7 | |
| 1999 | 50.2 | 1981 | 50.0 | |
| 1981 | 51.9 | 1982 | 50.0 | |
| 1994 | 52.3 | 1984 | 50.0 | |
| 1984 | 52.4 | 2002 | 50.0 | |
| 2002 | 52.7 | 1997 | 53.3 |
The start of 1983 might have had a good excuse for the chaos. Because of a 57-day players’ strike, the 1982 regular season was cut to nine games per team, and the postseason was also expanded from 10 to 16 teams. These unusual circumstances helped make the final, post-Super Bowl Elo ratings of 1982 among the most tightly bunched since the merger. Only the 1988 and 1981 seasons ended with a smaller standard deviation of team Elo ratings than 1982 did — which, in turn, helped set an uncertain tone for 1983 that didn’t let up all season. Across the entire regular season that year, Elo’s favorites ended up going 125-98-1, giving them their worst single-season winning percentage (56.0 percent) in any season since 1950, when our QB-adjusted Elo rating data begins.
1983 was a rough year for picking favorites
Lowest combined winning percentage for favorites (based on Elo forecasts) in a single NFL regular season, 1950-2021
| Favorites’ Record | ||||
|---|---|---|---|---|
| Season | Wins | Losses | Ties | Winning Pct |
| 1983 | 125 | 98 | 1 | .560 |
| 1956 | 40 | 30 | 2 | .569 |
| 1982 | 72 | 53 | 1 | .575 |
| 1957 | 41 | 30 | 1 | .576 |
| 1987 | 121 | 88 | 1 | .579 |
| 1955 | 41 | 28 | 3 | .590 |
| 2006 | 153 | 103 | 0 | .598 |
| 1994 | 134 | 90 | 0 | .598 |
| 1951 | 42 | 27 | 3 | .604 |
| 1981 | 136 | 87 | 1 | .609 |
We might be inclined to view 2021 through a similar lens, since the COVID-19 pandemic has us coming off one of the strangest years in sports history. Among 2020’s many peculiarities, NFL scoring was off the charts last year, and home-field advantage completely disappeared with reduced or eliminated crowds of fans for practically all season. NFL home teams actually lost more often than they won in the 2020 regular season — just the fourth time that’s ever happened in pro football history — going 127-128-1 on the year. Add in an eventful offseason, and you could see teams going into this year more scattered than usual.
Then again, most of the noise surrounding the NFL in 2020 didn’t actually seem to affect our model’s predictions. Looking back on our forecast data, we found that 2020 was (surprisingly) the model’s most accurate season since 2015, whether by judging how far off pregame predictions were or by how often Elo favorites won. Teams weren’t especially bunched-up in their ratings by season’s end, either, and the average team’s eventual end-of-season Elo was within +/- 68.8 rating points of where it started, which made 2020 roughly as predictable as any post-merger season (where the average error was +/- 67.2 Elo points). So some of 1983’s more unsettled elements are not necessarily repeating themselves now, in 2021.
Besides, not every shocking Week 1 portends a tumultuous season. In the other years when Week 1 saw favorites win less than half of their games — 1972, 1980 and 1999 — the most highly regarded teams rebounded just fine. Each of those seasons ultimately saw favorites win at least 64 percent of their games over the full schedule, which is pretty close to the 65.6 percent success rate for favorites overall since 1970.
It is true that Week 1 of a season can have bigger implications on the rest of the year than we might think, since its games carry a lot of information — both in terms of affecting the standings and giving us a first glimpse at how new rosters stack up relative to each other. And in that sense, the unexpected results of Week 1 did throw some teams’ playoff odds for a loop. The Football Team (down 29 percentage points), Packers (-19), Titans (-18), Falcons (-18), Ravens (-16), Bills (-13) and Browns (-10) all saw their chance of making the postseason drop by double-digits in the opening week,29 while the Saints (+25), Eagles (+22), Cardinals (+19), Steelers (+16), Dolphins (+14), Broncos (+14) and Seahawks (+10) all gained at least 10 percentage points of playoff odds.
Chances are that at least some of those teams will look back at the chaos of Week 1 as a sign of what was yet to come, for both good and bad. But with a larger sample of regular-season games to work with this year than ever before, which should (theoretically) help filter out fluke performances, it’s too early to say whether 2021 as a whole will be historically topsy-turvy — even if its opening salvo was nothing if not that.
Check out our latest NFL predictions.