≡ Menu

A Tale Of One Season

Let’s review the season of a mystery team from last year. This team had a pretty difficult schedule, but wound up with an average record. Here is how things broke down, starting with the good.

  • Mystery team played two home games against teams in the bottom quarter of the league (all team ratings in this post are using SRS). Those are the games where an average team should do well, and in fact, those were the only two games all year that the team won by double digits.
  • Mystery team had three other games where an average team would be “expected” to win based on strength of opponent and game location: Mystery team went 3-0 in those games, with an average margin of victory of 5.3 points.

But things were not so simple for our mystery team all year.

  • Mystery team played five games against teams in the top quarter of the NFL. The result? An 0-5 record, with an average margin of defeat of 16.4 points.
  • The remaining six games were ones where an average team would be “expected” to lose based on strength of opponent and game location, but were not against top-8 teams.  Mystery team had an average points differential of -3.7 in those games, and a 2-4 record.

To recap, Mystery team blew out the bad teams, beat the below-average teams, lost to the above-average teams, and was blown out by the great teams.  That, I think, is as unexciting as a season narrative can get.  But if a team goes 1-7 to start the season, and then 6-2 to finish, it’s easy to spin the “tale of two halves” narrative.

Yes, our Mystery team was the Detroit Lions.  Those three games against top-5 teams came against Seattle, Arizona, and Kansas City: those games were all in the first half of the schedule, and the Lions lost 13-10, 42-17, and 45-10.  It was in the second half when Detroit got to host Philadelphia and San Francisco, with the Lions winning 45-14 and 32-17. In the first half, Detroit played Minnesota twice and Denver; in the second half, the Lions faced Oakland, New Orleans, and the Rams.

Did the Lions improve from the first half to the second half of the season? Sure. The graph below shows how the Lions did relative to expectation, based on home field and the strength of the opponent. In general, the trend was positive — i.e., Detroit got better as the season progressed:

lions seasons

In the first half of the year, the only really good game the Lions played was in Seattle, and that one is arguably misleading because Seattle wasn’t playing very well that time of year, either. But in general, strength of schedule played a big role in Detroit’s resurgence: the three wins in December came against the Saints, 49ers, and Bears: New Orleans finished last in points allowed, San Francisco finished 32nd in points scored, and the Bears game was a week 17 contest that featured Marc Mariani and Josh Bellamy as Chicago’s starting wide receivers.

To be fair, Detroit nearly went 7-1 down the stretch, with one loss coming on a Hail Mary. The Lions were a competent — and borderling good — team for the last eight games of the season. But I can’t help but think a lot of the team’s Jeckyl and Hyde act was the result of Splits Happen and strength of schedule.

  • Trepur

    Could you post a trendline for the above graph? It’s clearly a postive trend, but would be nice to see the slope of said trend.

  • Wolverine

    I strongly believe that for the “middle class” of NFL teams, win-loss record in a particular year has more to do with schedule and injury luck, than it does with the roster quality, coaching, etc. The post-Millen Lions have been an essentially average team, that was lucky in some years, and unlucky in others. Good luck and 2-3 flukey plays in 2011 and 2014 led to double digit wins and playoff berths, while bad luck and 2-3 flukey plays in 2013 and 2015 led to 7-9 seasons and staying home for the postseason.

    However, it can’t be understated how putridly they played in the first half of 2015. I think “awful in the first half, competent in the second half” is fair summary of the 2015 Lions.

  • Josh Sanford

    One way of looking at this might be to look at their worst loss (-35) and their best win (+31) and add those together, for a total of 66. And then compare that to other teams. There are two potentially interesting data points there: the breadth of the point spread (obviously the 66 points is a little outrageous) but also the percentage of the spread that exists above zero. (i.e. a bad team might lose by 45 sometime, but it’s ‘best’ outing is a 3 point win–so that 92% of the spread is in negative territory. That could be an interesting query.