# The NFL Statistical Crystal Ball: What 2014’s Numbers Can Tell Us About 2015

Even with a 16-game sample, there’s a lot we can predict about the upcoming NFL season by looking at what just happened.

Each May, during the rare, glorious downtime of the NFL calendar, we take a look back at the underlying metrics from 2014 that will help predict what happens to teams in 2015. Many of them are familiar to regular Grantland readers, but here’s a useful primer on where they come from and how they work.

Of course, season-long statistics have a weaker predictive value in football than they do for just about every other professional sport, specifically because the season is much shorter. You can fit 10 NFL seasons into one MLB campaign with a couple of games to spare. The 16-game schedule means we learn less about the true talent level of each team in football than we do in other sports.

The good news is that we can turn that lack of information into a positive. Because 16 games just aren’t enough to learn much about a team, we can usually safely say that teams that exhibit some extreme characteristic or have some event occur a freakishly high (or low) amount of the time will not have that same experience over the next 16 games. That’s simple regression toward the mean. No stat is ever going to be a flawless predictor — again, we only get to see it bounce back in a 16-game sample — but history tells us that these figures give us some meaningful insight into the future before a ball is even snapped.

Let’s run through a few of these key statistics, see what they told us about 2014, and identify the teams that stand to improve or decline. It will be helpful to have last year’s version of this piece handy.

## Point Differential

Ah, the old standby. In every sport I’ve ever examined, the difference between the number of points a team scores and the points it allows is a better indicator of its future win-loss record than its actual win-loss record over that same time span. Pro football is no exception, where research by Daryl Morey (yes, the same Daryl Morey who is very sad about last night’s Rockets game) found that a variant of Bill James’s Pythagorean Expectation formula can estimate what each team’s win-loss record “should” have been, given their points scored and allowed.

This was a very useful piece of information last offseason. The team with the largest gap between its actual 2013 win-loss record and its expected win-loss record was the Texans, who went 2-14 despite possessing the point differential of a 4.2-win team. (They’ll come up a lot when reviewing last year.) Houston promptly improved by seven wins in 2014, and while it still underperformed its Pythagorean Expectation by 0.8 wins, a 9-7 record isn’t shabby for a team that started Ryan Fitzpatrick, Ryan Mallett, and Case Keenum at quarterback. The Falcons, Browns, Lions, and Washington were also among the five teams with the largest negative gap between these two figures, and they each improved, although obviously to varying extents. Sorry, Washington fans.

The dramatic outlier on the other side was another team whose fans often deserve condolences. The Jets managed to go 8-8 in 2013 by virtue of game-saving penalties in a number of close contests, including the questionable late hit call on Lavonte David that extended their Week 1 win over the Bucs and the personal foul on Chris Jones that eventually led to a win over the Patriots. Their point differential, though, only produced a 5.4-win Pythagorean Expectation. And in 2014, the bottom fell out. The Jets didn’t perform all that differently, producing a 4.9-win point differential, but a 4-12 record got everybody fired. Flip the luck between the two seasons and Rex Ryan gets fired in 2013 while John Idzik keeps his job after 2014.

Others were able to stave off a decline. The Colts played like a 10-win team but managed to go 11-5 for the third consecutive season, increasing their point differential for the third year in a row. The Patriots also managed to stick around at 12-4 for another year, but the Giants and Broncos each declined by one win to finish out the five most likely candidates to decline in 2014. What about 2015?

The obvious candidate to take a step backward would be the Arizona Cardinals, who went 11-5 despite outscoring their opposition by a mere 11 points, which paints them as a team more likely to be narrowly over .500. Since the league went to a 16-game slate, just one other team has managed to win 11 games while outscoring their opponents by fewer than 20 points: the 2004 Falcons, who went just 8-8 the following year.

Of course, the 2014 Cardinals weren’t your typical football team, which is what makes their 2015 season so interesting to project. Is it really fair to use their 2014 performance as a baseline when they got eight starts from Drew Stanton and two from Ryan Lindley? Throw out Lindley’s starts, and Arizona’s point differential improves to 43 points; then again, given that it lost both of those games, its 11-3 record with Stanton and Carson Palmer outpaced its expected record by … 2.7 wins. There will be plenty more on Arizona as we get closer to the season.

The Lions are another interesting case. They underperformed their Pythagorean Expectation in 2012 and 2013 because of their awful record in close games, going a grisly 6-14 in contests decided by seven points or fewer. Their underlying performance didn’t change that much from 2013 (8.5 expected wins) to 2014 (9.2 expected wins), but by flipping their luck in those one-score games to a 6-1 mark, Matthew Stafford & Co. pulled out an 11-5 season. It’s tempting to attribute that to replacing Jim Schwartz with a better late-game decision-maker until you remember that the guy they hired was Jim Caldwell. At least he won the Morning Bowl.

Actually, before I get to the team at the top of that other table and claw my brains out, let’s talk about how teams did in those close games.

## Record in One-Score Games

This is thee Pythagorean Expectation’s more tangible cousin. Point differential isn’t exactly complex, but it’s even easier to count up a team’s wins and losses in games decided by seven points or fewer and see whether they were able to produce a similar record the following year. The vast, vast majority of teams win neither a particularly high nor a particularly low percentage of their close games on an annual basis. Most of the differences between teams comes in what they do in the other games, the ones that are decided by two scores or more.1

1.

Yes, technically, an eight-point game is a one-score game. But for the purposes of having comparable data before 1994, when the NFL re-implemented the two-point conversion, I’ve stuck with the definition of seven points or fewer. Going with an eight-point definition would change some of the records here and there, but it wouldn’t materially change the concept.

Nobody shook their fist harder at everything I wrote in the previous paragraph than the Indianapolis Colts, who have been so far ahead of the curve that I devoted an entire article to their impudent ways during Andrew Luck Week last offseason. Their 14-2 record in one-score games during Luck’s first two years at the helm was virtually unprecedented.

Then, in 2014, the Colts started 0-2, losing their first two games by three (Eagles, 30-27) and seven (Broncos, 31-24) points. Problem solved, right? Well, not really. They went 4-0 in one-score games the rest of the way, including a pair of victories over the Texans that ended up deciding the AFC South. There was no reason to think Luck was going to fall victim to the gambler’s fallacy and that the Colts were going 0-6 in one-score games or something, but even a 50-50 split would be nice one year.

The league’s other outliers mostly bent to the will of time. The four other teams listed last year that did particularly well in one-score games were the Jets, Panthers, Patriots, and Eagles. They went a combined 20-8 in one-score games in 2013; one year later, they went 12-12-1, only saved from a losing record by a shanked Mike Nugent field goal. Throw in the Colts and they make it to 16-14-1 (.532).

And the five teams we would have expected to improve all did so in 2014. The Lions, who improved their close-games record from 3-6 in 2013 to 6-1 in 2014, were joined by the Texans (a brutal 2-9 in 2013), Steelers, Falcons, and Washington. Those five went 12-33 in one-score games in 2013 before improving to … 17-15 (.531), a 2014 win percentage virtually identical to those teams that were dominant in close games the previous year. The Falcons and Texans still weren’t very good in close games, each going 2-4, but the Lions and Steelers pushed this group over .500.

So, who are the obvious candidates to improve and decline this year? Here’s your last chance to jump off a very rickety bandwagon:

And there are the Buccaneers, standing out after going 1-8 in one-score games last season. That’s just unreal, and it’s how a 2-14 team that lost by 42 points to the Falcons still managed to post 4.5 expected wins. You would figure that some of those games were just meaningless matchups where the Buccaneers were down all game and managed to come up with some garbage-time drives from Josh McCown to make it close. That’s really not the case. The Buccaneers either had a late fourth-quarter lead and blew it or had a drive with a meaningful chance of winning in just about every one of these games!

That’s unreal. The second Panthers game is really the only one where the Buccaneers weren’t one drive away from winning.

There’s been hype for the Buccaneers on this site in years past. Robert Mays and I both backed them in 2013, and while I hopped off the bandwagon last year, Mays stuck around for another run. I wouldn’t blame anybody for ignoring the numbers with Tampa Bay, but the numbers have never been stronger in favor of their improvement. The 2013 team also had a staph infection run through the locker room during camp, ending two players’ careers, while its starting quarterback basically washed out of the league in entirely inexplicable fashion in a matter of weeks. That stuff seems unlikely to happen here.

Even beyond the numbers, it’s worth pointing out that the biggest year-to-year improvements (like the 2012 Colts and 2013 Chiefs) come when teams swap out replacement-level quarterbacks for competent-or-better passers. Even if you’re skeptical of Jameis Winston as a pro passer, it’s hard to imagine he’ll be worse than the combination of McCown and Glennon. I say this with trepidation in my heart: The Tampa Bay bandwagon is back and open for business.

On the other side, it’s unlikely the Packers will again finish undefeated in close games. It’s tempting to chalk up last year’s performance to Aaron Rodgers, ignore the numbers, and move along, but remember that the Packers were 5-1 in one-score games during that 15-1 season in 2011. The following year, even with Rodgers staying healthy, they went 3-3 in those same one-score games. I’d also be worried if I were a 49ers fan, given that six of their eight wins were of the seven-points-or-fewer variety. That Week 1 blowout of the Cowboys sure looked a lot different for everybody involved by the end of the year.

## Turnover Margin

As important as turnovers can be, teams on the extremes of the turnover margins find it very difficult to stay there. It’s just too difficult to maintain dominant performances on both sides of the football, regardless of how good you might be. Even if you carry the same personnel and play the same style, the randomness inherent in a 16-game schedule just makes it too tough to repeat.

In some cases, that can account for almost total randomness. Take the Texans, who had the league’s seventh-best turnover margin (plus-12) in 2012 before falling to the league’s worst mark (minus-20) with much of the same personnel in 2013. One year later, having replaced Matt Schaub with, um, Ryan Fitzpatrick, Houston’s run-heavy scheme bounded back toward the top of the class; its plus-12 mark was tied for the second-best rate in football.

Other teams saw their notable improvements strike plexiglass and bounce backward in 2014. The Chiefs and Eagles made historic jumps by moving from the very bottom of the 2012 charts to the top of the 2013 turnover leaderboard, but they each fell off. The Chiefs went from a plus-18 mark in 2013 to minus-4 last year, while the Eagles had Nick Foles turn back into a pumpkin in a year when they provided more giveaways (36) than anybody else in football. Their minus-8 mark was 26th in the NFL.

The only team really able to fight off the rot was Seattle, but even its dominance took a dip. After posting a league-high plus-20 turnover ratio in 2013, the Seahawks dropped to fourth-best in the league at plus-9 last year. To be fair, Seattle did manage to win the turnover battle by a lone takeaway during the postseason, but … well, I’m traveling to Seattle soon, so let’s just move on for my own safety. Who stands out as the teams to watch for in 2015?

It’s not good to see the Packers showing up again as a team with some likely regression coming. Tramon Williams wasn’t a fantastic cornerback by the end of his run in Green Bay, but since breaking into the starting lineup for good during 2008, only Charles Woodson, Ed Reed, and Asante Samuel picked off more passes than Williams, now a member of the Cleveland Browns. And far be it from me to expect anything but greatness from Rodgers, but his 1.0 percent interception rate from last year is totally out of line with his previous career rate of 1.8 percent; over the course of a full season, that would mean another five picks.

The Patriots can expect some decline if only by virtue of starting Jimmy Garoppolo for four games, while the Cardinals were somehow lucky in one way last year. Drew Stanton threw more dropped interceptions than anybody else in football. Don’t expect Arizona to be quite as lucky if it’s stuck going to Stanton again in 2015.

It seems too easy to just write off Oakland and any chance it will make a significant improvement, but improving its turnover ratio would be a way to take some modest steps forward. Then again, if Derek Carr’s finger injury is serious, Oakland would be forced to turn the offense over to Christian Ponder or Matt McGloin, which would be disconcerting. The Saints, who invested heavily in defense and seem to be moving toward a run-heavy approach on offense, are a more likely candidate for rapid improvement.

## Defensive Touchdowns Allowed

Let’s run one more stat out there. There’s no year-to-year relationship when examining the number of points a team allows on safeties and fumble and interception returns. Even teams that throw a lot of picks or fumble away a lot of footballs manage to stumble into a tackle or have defenders fall over to even things out.

Take last year’s leader in points allowed, the Bears, who started Jay Cutler for 15 games and lived to tell the tale. Even as the combination of Cutler and Josh McCown enjoyed a dynamite season in 2013, Marc Trestman’s bunch allowed a staggering 44 points directly on turnovers. The Bears actually turned the ball over more frequently in 2014 (29 giveaways to 23), but even with Cutler sacrificing footballs to the opposition on a team that had given up on its coach by December, Chicago only allowed two defensive scores for a total of 12 points in 2014.

Yes, 44 points is a lot, but somehow, there’s one team that managed to go even higher in 2014. Who propped fantasy football defenses up last year?

Oh my. That’s eight defensive touchdowns allowed by the Rams, which doesn’t even include the 99-yard kick return they allowed to Knile Davis. Want to give your team a frustrating, unfair stat with no possible route to improvement besides variance, Jeff Fisher? The Rams were 0-8 in the games where they gave up a return touchdown last year (including that Chiefs game) and 6-2 in the ones where they didn’t cough up a TD. That stat isn’t super meaningful — any record is going to look worse when you put a team down by at least one touchdown as part of the split — but it’s instructive in thinking about just how self-destructive St. Louis was last season.

Elsewhere, three teams managed to make it through the entire season without allowing any points on offense!2 Nary a single team managed to do that in 2013. The Chiefs win the consistency award, with their hyper-conservative offense having allowed only a single safety to opposing defenses in both 2013 and 2014. They technically allowed a fumble return for a touchdown during the 2013 playoffs, but of course, that was by Andrew Luck.

## In All …

2.

The Ravens get an asterisk, as they went the entire regular season with zero, only to give up a safety during the wild-card round.

There are plenty of other numbers to run through. I already looked at one, projected strength of schedule, last month. There’s defensive scoring and injury rates and fumble recovery percentages, and it’ll all come up as we preview the 2015 season over the next few months.

For now, though, this is some good insight into which teams might take steps forward or backward in 2015. As always, the steps are relative. The Packers are still going to be good, even if they win half of their close games and fail to post the league’s best turnover ratio. The Buccaneers aren’t going to be Super Bowl contenders, even if they realize their potential and have a far luckier 2015 campaign with Winston at the helm.

Then again, these are the sort of fissures that lead to bigger stories. The Packers aren’t going to suck, but if they drop from 12-4 to 10-6 and the Lions fall back toward .500, that opens up the NFC North for the Vikings to possibly leap into contention. If the Bucs go 8-8, that makes it harder for the Falcons and Saints to jump back into the playoff picture. And if the Cardinals and 49ers both decline, the Seahawks and Rams could parlay a weaker division into a pair of playoff berths. We won’t know what will happen until it happens, but by looking closer at 2014, it’s suddenly easier to narrow down a few plausible possibilities.