In my previous post, I discussed the use of ELO to measure the relative strength of National Football League teams over history, and how we can identify big upsets that have occurred over the years. Today I’ll wrap up with a few random observations about the data.
I mentioned that biggest upsets have happened in the past 30-40 years. For completeness’ sake, let’s not forget those earlier decades. The following table lists the biggest regular season and playoff upsets for the 50s, 60s, and 70s:
| Season | Favorite (ELO) | Underdog (ELO) | Favorite Expected Win% | Result | |
| 1950s Regular Season | 1950 | Rams (1562) | Colts (1312) | 86.0 | Colts, 22-21 |
| 1950s Playoffs | 1951 | Browns (1769) | Rams (1614) | 62.6 | Rams, 24-17 |
| 1960s Regular Season | 1966 | Patriots (1542) | Broncos (1267) | 87.6 | Broncos, 17-10 |
| 1960s Playoffs | 1968 (SB) | Colts (1803) | Jets (1644) | 71.4 | Jets, 16-7 |
| 1970s Regular Season | 1974 | Bengals (1623) | Oilers (1289) | 90.9 | Oilers, 34-21 |
| 1970s Playoffs | 1970 | Vikings (1737) | 49ers (1611) | 75.0 | 49ers, 17-14 |
Other than Joe Namath’s big upset in Super Bowl III, most of these games are long forgotten. That 1974 game between the Bengals and Oilers is interesting. Both teams finished the season 7-7, so you wonder why their ELOs were so different. But at the time the game was played, the Bengals were 4-2 (and coming off a playoff season the year before). The Oilers? Not only were they just 1-5, but in each of the two previous seasons they went 1-13. That’s three wins in their last 34 games. They were basically the worst team in the league for two years running, by far. When they won, it was a stunner.
What about the biggest non-upset in history? In other words, what game did ELO think was the most lopsided, and it turned out that way? It was played December 5, 1976, between the Pittsburgh Steelers (8-4) and the Tampa Bay Buccaneers. Pittsburgh was the two-time defending Super Bowl champs. After a slow start in 1976, there were on a roll, winning their last seven games by a combined score of 171 – 28. The Buccaneers were an expansion team, 0-12. The teams’ ELO ratings were 1717 and 1175, respectively. ELO thought the Steelers had a 97.1% chance to win. The Oddsmakers favored the Steelers by 27 points. It turned out pretty much as expected. Pittsburgh won 42-0.
We can also identify the “best” game ever played, in terms of team quality. The larger the ELO the better the team, so if the average of the ELOs of the two teams is high, that means you have two quality teams playing. The game with the highest combined ELO? The 2007 AFC Championship game, between the undefeated New England Patriots and the 13-5 San Diego Chargers, who were on an eight-game winning streak themselves. It’s a shame the Chargers weren’t at full strength, as some of their key players were injured. But of course ELO didn’t know about that.
And what about the “worst” game ever, where the two team had the lowest combined ELO? Ignoring the early days of the American Football League, and expansion teams (who automatically start with a very low ELO), the lousiest NFL game happened just last year: the New York Jets (ELO of 1292) vs. the Jacksonville Jaguars (1233). The Jets were 3-11, coming off a 2-14 season the year before. The Jags were 2-12, having been 1-15 the previous season. That’s a grand total of eight wins in 60 games, between them; no wonder ELO didn’t like either of them. The Jets won, 26-21. Well, someone had to.
