Continuing a tradition that we began last year, we've taken a data-driven look at which NFL head coaches are sitting on the hottest seats heading into the 2018 season.
Keep in mind we are not predicting which coaches will be fired. That involves two separate factors: How hot each seat is and the fortunes of each team in 2018. Instead, we isolated the first factor -- the heat under each seat -- by assuming they will each have the same losing season. Specifically, we presupposed every team would finish the year 4-12, and then asked our analytical model to make a projection on the fate of each head coach.
Just as before, we trained our model using Pro Football Reference's data on all NFL head coaching tenures since 1979. Unsurprisingly, the most important factor is the team's record from the present season, but there are a number of other factors that interact with each other within the model. Click here to read more on our methodology.
We've tested the model against every season for the past 39 years, and historically, it's correct 80 percent of the time, when using a cutoff of more than 50 percent predicting a firing, and less than 50 percent predicting retention. Last season was slightly unusual in that two of the five most likely coaches to be replaced, based on actual season results, survived. Hue Jackson (75 percent) and Bill O'Brien (65 percent) were both vulnerable, but were given reprieves due to extenuating circumstances. The other three, Chuck Pagano, Ben McAdoo and John Fox, were all replaced. Jack Del Rio's dismissal was thought to be unlikely by our model, but the Raiders jettisoned him to make a play to bring back Jon Gruden. Mike Mularkey was dispatched in one of those "mutual agreement" situations despite leading his team to a playoff appearance. As our model pointed out a year ago, he was on very thin ice before the season started.
Here's a look at which seats would be hottest assuming a four-win season in 2018:

1. Marvin Lewis, Cincinnati Bengals
Chances of being fired after a 4-12 season: 75 percent
Maybe there should be an extra variable in our model that just says, "Is it Marvin Lewis?" and it simply predicts he'll be retained 100 percent of the time. Lewis has been in place for 16 seasons now, second in tenure length only to Bill Belichick's 19, but they could not be more different in terms of postseason success.
Lewis' Bengals last made the postseason two years ago, and have never notched a postseason victory. Perhaps to the organization's credit, it has valued Lewis' overall winning record and his string of five winning seasons from 2011 to 2015, rather than reacting to misfortune in the playoffs. The Bengals have also been notably averse to getting onto the carousel of paying out the guaranteed contracts of fired coaches only to hire another coach who might soon be on the hot seat himself. But a 4-12 season would mean three straight losing campaigns in Cincinnati, something few coaches survive.

2. Todd Bowles, New York Jets
Chances of being fired after a 4-12 season: 71 percent
Bowles survived in 2017 because the Jets' five wins exceeded even lower expectations going into the year. Another down year, however, would mark his third consecutive losing season and things would be trending in the wrong direction. One thing in his favor is that he is somewhat tied to the hip with his GM, Mike Maccagnan, who hired Bowles before the 2015 season. The development of rookie quarterback Sam Darnold will be instrumental in Bowles' fortunes in New York.

3. Ron Rivera, Carolina Panthers
Chances of being fired after a 4-12 season: 70 percent
This may seem high to you, and it does to me too. But a four-win 2018 would be a precipitous drop from the Panthers' 11-win regular season a year ago, and would comprise a relatively poor record (21-27) since his Super Bowl appearance in 2015, which was the last season he won a playoff game. In Rivera's favor is the fact that GM Marty Hurney is back, and the two have a relationship from Hurney's first tenure with the team. But expectations are high in Carolina, so assuming there's no extenuating circumstances like an injury to quarterback Cam Newton, it's reasonable to think Rivera's seat could be fairly warm.

4. Jay Gruden, Washington Redskins
Chances of being fired after a 4-12 season: 69 percent
This would, hypothetically, be Gruden's third consecutive season of declining win totals in Washington. With just one winning season (9-7) that led to a playoff berth, his past four years have been marked by mediocre team records. The front office remains the same as the one that brought him on board; otherwise his seat would be even hotter.

5. Hue Jackson, Cleveland Browns
Chances of being fired after a 4-12 season: 68 percent
To some readers' surprise, Jackson did not make the top of this list last year following a 1-15 record in 2016. He, in fact, survived last season and was granted a "reset" by the owner and first-year GM John Dorsey. But another dreadful season for the Browns would certainly turn up the temperature under Jackson's chair. Reset or not, it's difficult to find many parallels in the data that would suggest he would get another opportunity to helm the Browns. Sometimes in these situations, even if a team's poor performance can't be laid at the feet of a coach, the stink of losing can be contagious.

6. Bill O'Brien, Houston Texans
Chances of being fired after a 4-12 season: 67 percent
O'Brien had modest success right away in his four-year tenure, notching three straight seasons with nine wins, in addition to two playoff appearances and a playoff victory. Last year, it appeared the Texans had caught lightning in a bottle with rookie quarterback Deshaun Watson, but they lost him to injury after six starts. That, together with a season-ending injury to defender J.J. Watt, removed O'Brien from the hot seat conversation. But a fifth year for O'Brien that features just four wins would definitely put the pressure back on. Our model is actually a combination of two different machine-learning approaches, and O'Brien is the coach for whom the two sub-models most disagree (49 percent versus 85 percent). The confusion may be due to dodging a firing after last year's four-win effort. Not many coaches survive that to begin with, so O'Brien's reprieve due to key injuries in 2017 makes this season a bit more uncertain.

7. Jason Garrett, Dallas Cowboys
Chances of being fired after a 4-12 season: 64 percent
Expectations are always high in Dallas, but following up a disappointing 2017 with a four-win season in 2018 would mean consecutive years of significant decline. That spells peril for a head coach in any town. Despite an overall winning record in his eight seasons as head coach for the Cowboys, Garrett has enjoyed only two notably strong seasons -- 12 wins in 2014 and 13 in 2016 -- and has just one playoff win. Injuries (and suspensions) have played a role in that record, but frustrations might finally boil over without significant progress in 2018.

8. Dirk Koetter, Tampa Bay Buccaneers
Chances of being fired after a 4-12 season: 64 percent
Similar to the situation in Dallas, a down year for Tampa Bay would mean a second consecutive season of decline, following Koetter's initial season, which produced nine wins (but no playoff berth). Without improvement for the Buccaneers in 2018, someone else is likely to be handed the reins in 2019.
Lukewarm seats
No coach is completely safe, especially after a down season for his team. Most coaches not noted above are in the 40 percent to 60 percent range, according to our exercise. Andy Reid in Kansas City, Mike McCarthy in Green Bay, and John Harbaugh in Baltimore are notable head coaches who would be at the top of that range. They each have distinguished résumés, but may want to consider updating them in the event of a disappointing year in 2018.
The next most vulnerable group tends to be all the first- and second-year head coaches. The safest coaches are the ones you'd expect -- Super Bowl winner Doug Pederson, Sean Payton (who reset his clock with a strong 2017), Bill Belichick and Mike Tomlin. Even that group is given a better than 20 percent chance by our model. It may seem crazy, but there are more than a few examples of coaches fired with relatively stellar records. Two-time Super Bowl champion Jimmy Johnson of the Cowboys might be the most notable, but there are others such as Marty Schottenheimer, who was fired immediately following 10- and 14-win seasons. George Seifert won a Super Bowl leading the 49ers, but was let go just two years later -- immediately after a 12-win season. More recently, John Fox was shown the door in Denver after averaging 11.5 wins in his four seasons there.
As always, every case is unique, and there are important aspects of each that even a sophisticated mathematical model can't consider. You can think of this exercise as a measure of how unexpected or unprecedented it might be to replace a head coach given his résumé.
Methodology
While the team's record is obviously of critical importance to whether a coach is fired, here are the other factors taken into account by our model:
The degree of decline or improvement from the previous season -- this is almost as important in our model as this year's record. The fact that GMs and owners are much more likely to pull the trigger after a team fails to meet expectations is hardly surprising, but it's important to note that a coach who goes from 3-13 to 5-11 is less likely to be fired than a coach who slips from 9-7 to 5-11.
The next most important factor is the coach's long-term record with the team, represented by his rolling average winning percentage for the past two through five seasons.
Playoff success matters. The model looks at how long it has been since a coach made the playoffs, and the number of seasons since a playoff victory. The recency of a playoff appearance appears slightly more important than that of a playoff win.
Did the current GM hire you? The thinking here is that a general manager (or whoever has the hiring and firing authority) is more likely to cut ties with his predecessor's coaching hire. Coaches generally have a 25 percent chance of getting fired when working with a GM who didn't hire him, as opposed to 18 percent when the current GM is the one who did.
Tenure has an impact on the model, but not in a straightforward way. Whether you've been the coach for a long time or you just got there, the data still relies on a combination of other factors to determine whether your seat should be hotter.
I also included Super Bowl appearances/victories as factors, but they turned out to be the least important. Coaches with a Super Bowl appearance or victory on their résumé aren't given much extra slack.