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:

Jump to a team:
DAL | LAC | CIN | DET

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.

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.

2019 point differential: minus-8
Pythagorean expectation: 7.8 wins
Record in games decided by seven points or fewer: 2-9
FPI projected strength of schedule: 16th toughest

The Chargers, of course, are also familiar to readers of this column. They were among the most likely teams to improve in 2016, when they jumped by one win. They made it back in 2017, and improved by two wins, and then jumped back on a third time in 2018, when they went from 9-7 to 12-4. That was enough to get them on last year's decline list, where they fell from 12-4 to 5-11. There's no such thing as a normal season for the Chargers.

The 2018 season was the only time this team seemed to show any propensity for winning close games; they went 5-1 in one-score games that year and a combined 9-29 in those same games in 2015, 2016, 2017 and 2019. They were 2-9 in one-score contests last season for an 11-game swing in plus-minus that matches that of the Cowboys.

Of course, nobody loses close games quite like the Chargers. A normal team would be crestfallen if it lost three games in the final two minutes in which its quarterback threw an interception in the red zone, as Philip Rivers did a season ago. But that's just a normal year for the Chargers. They went through two of the worst losses for any team in 2019:

• In Week 7, the Chargers trailed 23-20 and drove to the Tennessee 1-yard line after a pass interference call in the end zone. As long as it didn't turn the ball over, Los Angeles would get at least two shots to win the game and would be able to attempt a chip-shot field goal to send the game to overtime. Melvin Gordon appeared to score on first down, only for the franchise back to lose the ball on the way down, recovering it short of the goal line. On second down, Gordon fumbled again, with the Titans recovering for a brutal Chargers loss.

• Six weeks later, the Chargers battled back in the fourth quarter to tie the game twice against the Broncos. Keenan Allen was stuffed on a third-and-1 catch for no gain, and with 19 seconds left, coach Anthony Lynn decided to kick a field goal to tie the game at 20. The game seemed likely to go to overtime, but in a situation in which many teams would have kneeled and gone to OT, the Broncos sent Drew Lock back onto the field and had him throw up a 37-yard jump ball to Courtland Sutton. With the ball nowhere near Sutton, the only way Casey Hayward could have extended the game was by committing pass interference, which is exactly what the veteran corner did. The Broncos kicked a game-winning field goal on the next play.

Rooting for the Chargers is not for the weak of heart. Los Angeles lost three fumbles on plays from the opposing team's 1-yard line a year ago. The rest of the NFL lost one such fumble combined. No team had lost three of those fumbles in a season since at least 2001, which is as far back as ESPN's database goes.

I'd love to tell you that the close games were the only problem ailing Los Angeles a year ago, but it wouldn't be true. Owing in part to Rivers' 20 interceptions, the Chargers turned the ball over 31 times. That wouldn't be a huge problem if the defense forced a bunch of turnovers, but Gus Bradley's unit was able to generate only 14 takeaways, the fewest of any team.

Much of that owed to a spectacularly bad fumble recovery rate. Defenses typically recover about 48% of the fumbles they force, depending on how and where they happen. The Chargers forced 16 fumbles last season, which should typically produce about eight giveaways. They instead recovered just three. While the Dolphins were actually slightly worse a year ago, Los Angeles produced the fourth-worst fumble recovery rate for any defense since 1991. Every piece of evidence we have suggests that this amounts to nothing more than blind luck or randomness. Forcing fumbles might be a skill, but once they hit the ground, anything goes.

In all, the Chargers' turnover differential was minus-17, which tied them with the Giants for the worst mark in football. We all know winning the turnover battle helps you win football games, but typically teams with dramatic turnover differentials don't have those same issues the following year. That should work in the Chargers' favor.

To prove the point, if we go back through 1989 and look at every team with a turnover differential between minus-20 and minus-15, their average turnover differential the following season was ... plus-0.3. Those 44 teams regressed almost exactly to the mean. Unsurprisingly, their record also improved; they won an average of 2.5 more games than they had the prior season. The only team to fit in this category for last year's column were the Bucs, and even with Jameis Winston at quarterback, they improved their turnover margin from minus-18 to minus-13 when the defense improved.

The Chargers should also benefit by being healthier. While their struggles with injuries have been a seemingly annual problem, their 92 adjusted games lost was their highest total since 2016. Their two star offensive linemen in 2018 were Russell Okung and Mike Pouncey, who didn't take a single snap together because of injuries. On defense, they lost star safety Derwin James for most of the season after he injured a foot in training camp. Players such as Melvin Ingram, Brandon Mebane and Denzel Perryman all missed time.

Strangely, a special-teams injury really hurt the Chargers. Kicker Michael Badgley finally solidified what had been a disastrous situation for this team in 2018, but while the Miami product was battling a groin injury in camp, they kept Badgley on the roster and didn't sign a replacement. At first, they used punter Ty Long on field goal tries before importing kicker Chase McLaughlin. Badgley spent eight weeks taking up a 53-man roster spot as an inactive before he was finally ready to kick.

He was 13-of-16 on field goals after returning, but his replacements went a combined 13-of-18 (72.2%) on field goal tries in his absence. It cost the Chargers dearly when Long missed two field goal attempts in a 13-10 loss to the Lions. The Chargers were lucky with opposing kickers, with misses from Adam Vinatieri and Eddy Pineiro leading to their two wins in close games a year ago, but they would expect to have a better kicking game in 2020.

One unexpected way the Chargers might benefit from the 2020 season: silence. They have spent the past three years in Carson, California, playing in front of a crowd that consisted mostly of the other team's fans. At times last year, they had to use a silent count at home because the other's teams fans were making too much noise.

The Chargers are moving to a new stadium in 2020, and while their fans might turn out to be outnumbered by an even greater extent in a much larger venue, it seems unlikely that they'll be playing games in front of any fans whatsoever this upcoming season because of the coronavirus pandemic. That will reduce the home-field advantage of teams that rely on their fans to create an intimidating atmosphere. For a Los Angeles team that might have been intimidated to play in its own stadium last season, silence could be golden.

My big concern with this prediction? Losing Rivers. When we use past data to predict how a team will play in the future, we're estimating that their overall level of team talent hasn't changed all that much. In most cases, even if they lose a key player at another position, that turns out to be true. Rosters change every year, but the factors I've mentioned in these previews still turn out to be useful predictors of future team performance.

When a team makes a major change at quarterback, on the other hand, it's more likely to impact the team's overall level of talent. Last year, both of the teams that failed to live up to our expectations changed quarterbacks, and while Rivers had his problems, the Chargers could still be downgrading under center. Tyrod Taylor avoided turnovers and was an effective low-end starter with the Bills, but he was a mess in his short stint with Cleveland. Rookie first-round pick Justin Herbert is coming into the league as a project; it's not impossible to imagine the sort of inconsistent play from him that we saw from Daniel Jones as a rookie.

With that in mind, while I think the Chargers will improve, I'm not as optimistic about their chances of returning to the postseason as I would have been if they had kept Rivers. A leap to the 7-9 or 8-8 range seems more plausible.

2019 point differential: minus-141
Pythagorean expectation: 4.4 wins
Record in games decided by seven points or fewer: 0-7
FPI projected strength of schedule: 13th easiest

If a team upgrades at quarterback, it's also going to help its chances of improving. The Bengals will presumably upgrade under center by replacing the combination of Andy Dalton and Ryan Finley with No. 1 overall pick Joe Burrow. The team with the worst record in football almost always improves the following season, and the past 10 times a team drafted a quarterback with the first overall pick, it improved by 3.6 wins the following year.

We've highlighted a number of those teams for improvement in years past, including the 2015 Bucs and the 2018 Browns, who combined to win 11.5 games more than they had the year before drafting their new quarterback.

Teams that have the first overall pick are almost always bad. They're also often unlucky. The Bengals, to start, were decimated by absences. A.J. Green and Jonah Williams, the team's best receiver and most important offensive lineman, each missed the entire season with injuries. They struggled with the fifth-most injured offense in football by adjusted games lost. Red zone performance is relatively random from year to year, and while the Bengals ranked 21st in red zone trips last season, they were the second-worst offense in the league in terms of points per red zone possession. (They were the league's second-best red zone offense in 2018.)

play

Mike Tannenbaum and Ryan Clark discuss their expectations for Joe Burrow in his rookie season with the Bengals.

As you might suspect from a 2-14 team, Cincinnati wasn't good in one-score games. It was 0-7 in games decided by seven points or fewer, and while a couple of those weren't as close as the final score indicated, the Bengals lost some legitimate heartbreakers. They were up 17-14 on both the Seahawks and Bills in the fourth quarter and lost both games. They were down 13-10 against the Steelers in the fourth quarter and drove into the red zone, only for wide receiver Tyler Boyd to fumble on the 6-yard line and hand the ball to their hated division rivals. They eventually lost 16-10.

The Bengals also saw a pair of late comebacks fall agonizingly short. Down 23-9 in the fourth quarter against the Cardinals, Andy Dalton led a pair of touchdown drives in the final five minutes to tie the game. The defense promptly allowed the Cardinals to drive back into field goal range in a 26-23 loss. In Week 16, Dalton did one better and led three touchdown drives in the final seven minutes, including two touchdowns and two two-point conversions in the final 35 seconds of regulation, to push a game against the Dolphins to overtime. The Bengals then somehow couldn't score on two overtime drives, with Miami kicking a field goal with four seconds left in OT to win.

One of the other problems with being the worst team in the league is that you don't get to play yourself. The Bengals faced a tough schedule in 2019; Football Outsiders pegged it as the league's sixth-toughest slate. This year, while the AFC North should be tough, the Bengals should play an easier set of opponents, as they'll swap out the NFC West for the NFC East and the AFC West for the AFC South.

Let's be realistic: The Bengals were not a good team. They were 29th in DVOA and got that far only because they had the league's best special teams. (Even there, though, opposing kickers hit 88.2% of their field goals against Cincy, the fifth-highest rate in the league.) Zac Taylor's group wasn't a good team with bad luck. It was a bad team with bad luck. In 2020, the Bengals can be a mediocre team with normal luck, and that would be enough to push them to 5-11 or 6-10.

2019 point differential: minus-82
Pythagorean expectation: 6.0 wins
Record in games decided by seven points or fewer: 3-7-1
FPI projected strength of schedule: fifth easiest

If you take a glance at the 2019 Lions, you probably think you would have most of the story for why they will improve in 2020. Detroit was 3-4-1 with Matthew Stafford under center, but after the 2009 No. 1 overall pick was sidelined by a back injury and missed the second half of the season, it went 0-8. Stafford is expected to be fully healthy for 2020, so if the Lions can get the Stafford who excelled before getting hurt, they should be much better in 2020. Right?

Well, that's part of why they are likely to improve, but it's not the whole story. The gap between the Stafford Lions and the Lions who were led by a pair of replacement-level quarterbacks in David Blough and Jeff Driskel wasn't quite as big as it seems by their record. The Stafford Lions were a little lucky, and the non-Stafford Lions were more competitive than their record indicated.

They both had one thing in common: They couldn't hold on to their leads in the fourth quarter. Detroit lost five games in which it held a lead inside of the final 15 minutes. No other NFL team last season lost three. In four of those games, it held a lead in the final seven minutes. No other NFL team lost two such games. Look away, Lions fans:

• In Week 1, they blew a 24-6 fourth-quarter lead against the Cardinals and ended up tying. We're not even counting that one in the loss column.

• Three weeks later, they were up four on the Chiefs late in the game, only for Patrick Mahomes to convert a fourth-and-8 with his legs and set up the game-winning touchdown with 23 seconds to go. Earlier in the game, the Lions fumbled the ball on the Kansas City 1-yard line, with Bashaud Breeland returning it for a 100-yard touchdown. That was, by expected points added, the worst play of 2019.

• In Week 6, the Packers kicked a field goal at the end of regulation to beat Detroit 23-22.

• After Stafford got hurt, Detroit held a 16-13 lead against Washington at the two-minute warning. The Lions allowed a game-tying field goal, Driskel threw a pick, and another field goal won it for Washington in regulation.

• On Thanksgiving against the Bears, the Lions were up 20-17 in the fourth, only to allow Mitchell Trubisky to lead a touchdown drive with 2:17 to go.

• In Week 17 against the Packers, the Lions led 20-13 after scoring a touchdown with 11:30 to go. The Packers tied it up with an Allen Lazard touchdown, and after a Lions punt, an unnecessary roughness penalty and a 31-yard catch by Aaron Jones set up another game-winning field goal for the Packers.

This doesn't even include the Raiders game in which the Lions tied it with 5:22 to go, allowed the Raiders to score a touchdown, and then drove all the way to the 1-yard line before failing on fourth-and-goal from the 1 with eight seconds to go. That fourth-down failure was Stafford's final play of the season, which feels fitting. The Lions did need some luck to pull out narrow victories over the Chargers and Eagles, but what happened to Matt Patricia's team in the fourth quarter of games last season was truly staggering.

To put this in context using ESPN's win expectancy metric, the Lions ranked 20th in win probability added through the first three quarters of games in 2019, falling between the Eagles and Raiders. In the fourth quarter and overtime, their performance was worth minus-4.2 wins, which is the worst mark of the past 10 years.

Here's a list of the 10 worst teams in the fourth quarter and OT over that time frame besides last year's Lions and what happened to each team's record the following season:

Those 10 teams improved by an average of 3.5 wins. (Expand the sample out to the top 30 teams and they improved by an average of 3.3 wins.) I wrote about many of them before the season as likely to improve, including the 2018 Colts, 2015 Buccaneers, 2017 Chargers, 2018 Browns and 2014 Texans.

Elsewhere in the top 20 for worst fourth-quarter performance, you'll find the 2016 Jaguars and 2018 49ers, both of whom improved dramatically in the final stanza and went on long playoff runs after being featured as teams likely to improve in this space. I'm not saying this to toot my own horn -- I also picked the 1-15 Browns to improve off of this list and that didn't happen -- but instead to point out that this is generally a very fruitful place to look for teams about to make dramatic turnarounds.

The FPI also thinks Detroit will be aided by a softer schedule. It is projected to have the fifth-easiest schedule in the league this season; while it has to play the NFC South, it will get four games against the AFC South and six against an NFC North division that looks better on paper than it does to the algorithm. Every schedule projection is subject to change given the nature of this upcoming season, but the Lions should have an easier road ahead.

They will be better in the fourth quarter in 2020. They'll get a healthier Stafford. They'll most likely have a healthier offense altogether, given that they ranked 29th in offensive adjusted games lost last season. I'm concerned about the propensity for players to leave Detroit and play better after getting away from Patricia, and that might impact the Lions' ceiling, but I do think they'll be a better football team this season. Their chances of competing for the NFC North title are better than you think, in part because of who and what we'll talk about tomorrow.