By Harrison Chase and Kurt Bullard
Today signals the end of the NFL regular season, and for some coaches, it may signal the end of their current coaching tenures. To that end, we figured now would be an opportune time to break out our coach firing model that we have used a few times this season. In addition to calculating the raw percentages of each coach getting fired, we can now—with only one game left —break down the probabilities of each coach getting fired with a win and a loss, and see who has the most to gain from their Week 17 game.
First, let us simply look at the probability that the remaining coaches will be fired.
By our model, there is clearly one team that has a much larger chance of firing their coach than others: the Cincinnati Bengals. At 5-9-1, Coach Marvin Lewis has his team performing far worse than in years past. Although the Bengals have made the playoffs for the past five years, and some may point to that as reason why he should stay, our model views ‘playoff success’ as making the divisional round, which, with five straight Wild Card games losses, Lewis has not accomplished once in his tenure with the Bengals. However, despite our model’s confidence that he will be fired, reports seem to indicate that Lewis will be back for the 2017 season.
Following Lewis are two coaches hovering around 50% – John Fox of the Chicago Bears and Mike McCoy of the San Diego Chargers. With our model viewing their firing as basically a flip of the coin, neither of them can feel too comfortable.
While most of the probabilities seem fairly reasonable, with the top coaches being the ones most linked to the hot seat and the bottom ones agreed on to be fairly safe, there are two in particular that stand out to me: one higher than the general consensus seems to be and one lower.
Bruce Arians: Sitting at 27%, seems rather high, as very few reports have him on the hot seat. What is the reason for this discrepancy? In his fourth season, he has run out of leeway with our model (which gives a slightly boost to coaches in their first three years), and he has only one divisional round appearance in those four years. Therefore, our model views him as fairly long tenured coach with limited playoff success and a bad winning percentage this year. However, it fails to capture the Cardinals impressive last three seasons and the fact that they made the conference championship last season, all facts that the general public may be basing his job security in. While it seems unlikely that the Cardinals would fire him after one bad season, his situation is one worth keeping an eye for next year.
Chuck Pagano: On the flip side of the coin, the Colts’ coach is one that our model thinks is fairly unlikely to be fired yet has been on the hot seat for basically a full year now. Comparing his case to that of Bruce Arians, there are few key differences. First, Pagano made the divisional round in two of five years, a rate nearly double that of Arians (40% vs 25%). Second, he already has 7 wins this year while Arians could end up with 6 – we found that the difference between 6 wins and 7 wins is the most significance difference of one win (and have a dummy variable in our model to account for that). Finally, and perhaps most significantly, the Colts had only 2 wins the season before Pagano took over, while the Cardinals had 5 before Arians, which our model looks at, and credits Pagano with improving the team more compared to where he inherited it from.
Looking at these two cases which do not seem to agree with public opinion offer some suggestions on how to improve the model. In particular, I think there are two key things that would be worth looking at:
- The relationship between success and years since said success: it seems that the public is quicker to forgive Arians, whose success was only one year ago, compared to Pagano, who had it two/three years ago.
- The relationship between performance compared to before you took over and years of tenure: it seems that the fact that the Cardinals had more than twice as many wins as the Colts before their respective coaches took over is now largely irrelevant, once we get four/five years into their tenure.
We do not know if there will be enough data points of this nature to back up the suggestions above, but it is definitely something we will look at.
Moving onto which coaches have the most to gain from the Week 17 game, below are the five coaches with the largest difference in their firing percentage with a win compared to with a loss.
Leading the charge is the the Cardinals, whose seventh win would push Bruce Arians above the aforementioned threshold and give him a huge boost. After that are all the teams most close to a 50% chance of firing their coach. This makes sense – our model is the most unsure about these teams and therefor a win matters the most. None of these other teams are currently on the edge between 6 and 7 wins, which is why it matters even more for Arians than for these other coaches.
We can also see by looking at this that McCoy and Fox would have a greater than 50% chance of being fired with a loss, and less than a 50% chance with a win. Therefore, it is safe to say that McCoy and Fox are coaching for their jobs this Sunday.
