One-armed Knife Sharpener
- Apr 7, 2013
Bill Barnwell ESPN Staff Writer
Each year around this time, I try to figure out what's going to happen in the upcoming NFL season. Doing that in August isn't exactly an exclusive hobby of mine, but I try to focus on using metrics and measures that have historically been good predictors of future team performance. What happened last year matters, but not always in the way you might think.
In recent years, my predictions have been pretty accurate. I've split this column into winners and losers over the past three years, and over that time frame, I've identified 16 teams that were likely to improve the following year. Twelve of them have done so, with the average team's record leaping by 3.3 wins. Ten of the 16 went over their preseason over/under win total as listed on Pro Football Reference. When we include the teams we've predicted to decline, this column has gone 26-6 in predicting win/loss direction and 22-10 versus over/unders over the past three seasons.
Three of the five candidates we selected to improve in 2019 did so. The New York Jets and Tampa Bay Buccaneers each added two wins to their ledger. The New York Giants took a step backward as they rebuilt around rookie quarterback Daniel Jones, while the Carolina Panthers never had a healthy Cam Newton and fell apart after a hot start. Losing your starting quarterback, as the Panthers did, is more meaningful than any predictor we can invent.
The big success story was the San Francisco 49ers, whom we pegged as a team that could "make an unlikely trip to the playoffs." By the end of the year, things hardly seemed unlikely; the Niners won the NFC West, were the top seed in the NFC and made it all the way to Super Bowl LIV. Over the past three years, this list has included a pair of unlikely Super Bowl participants out of the NFC in the 2019 49ers and the 2017 Philadelphia Eagles, who jumped from 7-9 to 13-3.
Let's get to the four teams most likely to improve in 2020 before hitting the four teams most likely to decline Tuesday. Given the uncertainty surrounding the upcoming season, everything included here makes the (unrealistic) assumption that this will be a normal campaign. There's no way to project a team's strength of schedule when we don't know whether they'll even complete a full season. If we get something resembling a normal season, these are the teams most likely to take a step forward in 2020:
2019 point differential: plus-113
Pythagorean expectation: 10.7 wins
Record in games decided by seven points or fewer: 0-5
FPI projected strength of schedule: 12th easiest
The Cowboys were on the other side of this list last year as one of the teams I expected to decline. They did, falling from 10-6 to 8-8, but it wasn't because their performance declined. On a snap-by-snap basis, they were better than they were in 2018. They outscored opponents by less than a point per game in 2018, with that mark jumping to more than a full touchdown per game last season. They improved from 21st in DVOA in 2018 to sixth in 2019. They finished the season seventh in ESPN's Football Power Index, just ahead of the two teams that lost in the conference championship games.
What changed is simple. In games that weren't decided by seven points or fewer, the 2018 Cowboys were 2-4. The 2019 Cowboys were 8-3. In the close games that were decided by seven points or fewer, though, they fell from 8-2 in 2018 to 0-5 last season. The same regression to or past the mean helped sink them in 2015 and 2017, and while I don't want to suggest there's a Bret Saberhagen thing happening here, it popped up again in 2019 and cost Jason Garrett his head-coaching job.
Ezekiel Elliott and the Cowboys have a loaded offense and should make a run at the NFC East title. Larry W. Smith/EPA
More than anything, the Cowboys were just fantastic when they needed to be in the final minutes of games in 2018. They kicked a field goal on the final play to win two games and beat the Giants with a touchdown and a two-point conversion with 1:12 to go in Week 17. They beat the Eagles with a late score and two defensive stops inside the final four minutes in Week 10 and then topped Philly in overtime with an Amari Cooper touchdown in Week 14.
Last year, with a better offense and virtually the same core of talent, the late-game heroics didn't show up. Dallas laid an egg against the Jets, and after scoring two touchdowns in the fourth quarter to get it to 24-22 with 47 seconds left, it failed on the 2-pointer to push the game to overtime. Down 28-24 against the Vikings, Dak Prescott went 6-of-7 for 79 yards to push the Cowboys into the red zone with 1:57 to go. Facing second-and-2 from the 11-yard line, they handed the ball to Ezekiel Elliott two times and saw their star back lose a total of 3 yards. Prescott threw an incompletion on fourth down and then an interception on a Hail Mary to end the game.
The Cowboys were plus-6 in one-score games in 2018 and minus-5 in those same games in 2019. That's an 11-game swing over the course of two seasons. Since 1989, just five other teams have dealt with an 11-win swing or more in close games, one of which will be appearing later in this column. To get something resembling a significant sample, we have to expand a bit and consider the teams that had a negative swing of eight games or more. When teams typically undergo that sort of swing from year to year, what happens in the third season?
They almost always improve. Of the 27 teams that fell off by eight or more wins in close games, 23 improved the following season, while one stayed at their prior record and only three declined. Three of the four teams that didn't improve either replaced their quarterback by choice or via injury, including last year's Panthers, who got only two injury-hampered games from Cam Newton. The 27 teams improved by an average of 2.7 wins the following year and won just over 46% of their close games. Dallas should be better in those one-score games in 2020.
To put it another way, let's also take the Cowboys' point differential of plus-113 in 2019 and expand it out to consider teams that outscored their opponents by a total of 100 to 125 points over a full season. Last year, the only other two teams in that group were the 13-3 Saints and the 10-6 Vikings, each of whom made the playoffs and posted a better record than last year's Cowboys.
Over the past 30 years, 52 other teams have landed in this 25-point bucket. They won an average of 11.2 games. Just one other team -- the 1989 Bengals -- failed to post a winning record. They also improved the following year (although it was by only one game).
With better luck, Dallas would project as one of the best teams in football, given that it was one of those teams a year ago. If anything, it wouldn't shock me if the Cowboys actually were a little worse on a play-by-play basis and still improved their record anyway. They ranked second in offensive DVOA in a season in which they were the second-healthiest team in the league by adjusted games lost. They will have to replace retired center Travis Frederick, although they likely upgraded in the slot by swapping out Randall Cobb for first-round pick CeeDee Lamb.
This sounds like a simple concept, and I'm sure longtime readers aren't hearing anything new when I say this, but the simple reality of the NFL is that the easiest way to find which teams are likely to improve or decline the following season is to look at their record in close games. There will always be exceptions, but the vast majority of the time you'll find that teams that either win or lose their one-score games at a drastic rate one season don't repeat that feat the following year. It was true for the 2019 Cowboys. I expect it will also be true for the 2020 Cowboys.
You can read the rest of the story at Are the Cowboys contenders? Barnwell predicts NFL teams most likely to improve.