We introduced our new metric, Expected Points
, for NFL games using simple box score statistics to predict expected scores and win percentages for each game. Now that the season has ended, we'll go through the standings and the biggest standouts for the '16-'17 season.
The clear cut favorite is the Patriots. For consistency, we threw out their first four games without Tom Brady under center. After a total of 12 games, the Patriots averaged a 69.0%
win percentage in the games that Brady started. New England's offensive efficiency certainly helped their defense, but the defensive side of the ball was an entire point better, per expected points, than the next best team.
The Falcons show up in second place at 62.9%
in average win percentage. The Falcons blew out a lot of teams this season on the scoreboard and in our model. They have the best offense in the NFL at an Expected Points value of 29.7 per game. Their defense is actually about league average at 21.7 expected points allowed per game, but again, that's slightly misleading because of their offense's ability to control the game.
After removing the last game of the season for obvious reasons, the Raiders come in third. Oakland finished fifth in offensive Expected Points, but those days are long gone without Derek Carr. Connor Cook will be getting the start and he is approximately a touchdown worse than Derek Carr against a league average defense. Oakland has a chance to win their Wild Card game against the drowning Texans' offense, a little less than 40% per Expected Points and the betting market, but their Super Bowl hopes are in great danger. Oakland finished the season with a -0.4 yards per play differential, but this model suggests that they may actually have been as good as their record indicated, after all.
Similarly, we removed Week 17 for the Cowboys who come in at a surprising fourth place with a 59.5%
average (although they can be considered third with Oakland clearly out of the running). Despite still being very good, the Cowboys outscored their opponents by two more points per game than expected. This is purely on the defensive side of the ball. The Cowboys are about league average in every defensive metric. However, when we look at the drive statistics, they may be a bit lucky. For example, they allow the third most plays per drive and the eighth most yards per drive, but have only allowed scores on 35.1% of drives, which is about league average.
On the opposite side, there were two terrible NFL teams this season, the Rams (30.6%)
and Browns (30.4%)
. The 49ers (32.9%)
and Jets (34.4%)
follow with a small gap between them and the Rams. After that, the Jaguars (40.7%)
come in fifth worst with a large gap between them and the Jets.
We mentioned a few weeks back that the Giants were the luckiest team in the last three years. However, with their better than normal play in the last two weeks of the season and actual unfortunate results (despite winning in Week 17), they regressed closer to expectation. The Giants finished this season outscoring their opponents by 1.6 points on average. Per Expected Points, their expectation would have been flipped and should have been outscored themselves by 1.5 points. That margin would have finished 23rd in the NFL in our expected margin of victory.
The difference for the Giants comes almost entirely on the defensive side. New York has allowed nearly 340 yards per game this season, rank 12th in turnovers per game, and somehow allowed less than 18 points per game. Again, Expected Points is backwards-looking. Thus, the current state of teams may not be considered. For example, the Giants had quite a few injuries in their secondary to start the season. If we simply start at Week 7, the Giants have an Expected Points against of just 19.5 points per game. Since Week 10, that number goes down to 18.3 average points against. Using arbitrary end points is typically a bad method; however, with the injuries the Giants endured and their last few weeks being a much better reflection of their current team, it may be a useful study. If we remove the first six weeks of the season where they were injury-plagued, the Giants jump all the way up to 11th in Expected Points win percentage at 56.3%
. This seems to be a much better reflection of their current team.
Lastly, the Cardinals are the most unfortunate team in the NFL this season. Per Expected Points, the Cardinals rank fifth with an average win percentage of 59.2%
. There was an interesting trait in the Cardinals this season that caused a boosted average. For example, Expected Points, based on their boxscore statistics, only projected the Cardinals as favorites in seven of the games they played this season, or the exact amount they ended up winning. Of those seven “wins,” they had five such games that they were projected to win more than 89% of the time. These games drastically drove up their average win percentage. The games they actually won back it up as they won five games by more than a touchdown.
Simply put, the Cardinals seemed to either play close games or blow teams out this season. If we use the median win percentage of Expected Points for their games, it's only 48.5%
. That's a massive difference compared to 59.2% above and would move them from 5th to around 20th in the standings. The question for Arizona going forward is whether or not they had some extremely unlucky point distributions or if they were just simply good enough to blow out bad teams.
All in all, it was fun to introduce another model this season, and one that we look forward to adding to in the offseason. There are still plenty of improvements to be made and we hope to continue to do that going forwards.