New head coaches improve their team’s record 63% of the time, and on average, improve their record by 1.3 wins. But is this 1st-year lift just a mirage?
by Jason Pauley
Over one-fifth of head coaches are replaced annually in the NFL on average. We see about seven new head coaches per year. More often than not, the teams that change their coach improve. From 1979–2017 (the years in my study), teams that replace their head coach improve their record 63% of the time in the first year with their new head coach. They only decline 28% of the time (their record remained the same 10% of the time). The average team with a new head coach increases their win total by 1.3 wins.
Going into last season (2020) there were five new hires.
Three of the five led their team to more wins than the previous season, one remained the same, and one had fewer wins. Mike McCarthy was the only coach to lead his team to fewer wins, but losing his Pro Bowl Quarterback early in the season didn’t help his cause.
This year there are seven new head coaches.
Historical data would suggest that 4 to 5 of these coaches will improve their team’s record this year. Those teams that improved will feel good about the early results they’re seeing from their new coaches. Just like nearly two-thirds of past coaching changes, their year one improvement may be seen as a causal result of their coaching hire.
- Robert Saleh, Jets
- Nick Sirianni, Eagles
- Urban Meyer, Jaguars
- Brandon Staley, Chargers
- Arthur Smith, Falcons
- David Culley, Texans
- Dan Campbell, Lions
It’s clear based on the data from my research that hiring new coaches provides a team with a boost in year one. Either their new ideas, new philosophy, play calling, or removing an underperformer led to an improvement 63% of the time. If I stop writing right now, this seems to be a reasonable conclusion.
But let’s take another look at the 2021 hires, this time with their team’s record in 2020:
- Urban Meyer, Jaguars (1–15)
- Robert Saleh, Jets (2–14)
- Arthur Smith, Falcons (4–12)
- David Culley, Texans (4–12)
- Nick Sirianni, Eagles (4–11–1)
- Dan Campbell, Lions (5–11)
- Brandon Staley, Chargers (7–9)
It’s safe to say that there are two coaches on this list who are almost guaranteed to improve their teams’ record. Urban Meyer and Robert Saleh have almost nowhere to go but up. 90% of 1-win teams in my study (n=10) improve the next year (the 2016–2017 Browns went from 1–15 to 0–16). 2-win teams improve 86% of the time (N=29). We have a handful of 4-win teams in this group and those teams also improve at a high rate of 85% (N=78).
It’s a fact that hiring a new coach is followed by a win improvement of +1.3 wins. It’s also a fact that 63% of the time a team gets better. But when do teams hire new coaches? They hire them when they hit rock-bottom. The teams that hire new coaches are usually really bad teams. The average number of wins for a team that hires a new head coach is 5.7 wins. 61% of teams with three or fewer wins will hire a new coach the next year (N=76). 37% of teams with 4–6 wins hire a new coach (N=258). And of course, teams with a winning record (9+ wins) don’t often change their head coach; this only happens 8% of the time (N=473).
If you look at the chart below, you’ll see that teams at the very bottom of the standings tend to improve and teams at the very top of the standings tend to decline. This, as many football fans have come to understand, is regression to the mean.
The worst teams in the NFL are the worst for two reasons:
- They’re bad
- They’re unlucky
Reason #1 is often repeatable. If most of your team is bad at football in year N (relative to the rest of the league), then you will likely have a bad team in year N+1. Players will get better (or worse), the roster will change (but not all), but many of the underperforming parts will remain. What isn’t predictable or as repeatable is reason #2, luck and randomness. Luck and randomness are even stronger influences in a small sample size like 16 games (17 games starting in 2021). How a team performs in 1-score games, fumble luck, schedule strength, and injury luck are all factors that are less repeatable, and more luck-driven. A 2–14 team is probably not 2–14 bad, they needed some bad luck along the way. So, they will have positive regression the next year, with or without a new coach.
We already know the worst of the worst teams are the teams that hire new coaches, so is it the act of hiring a new coach that improves teams records in year N+1, or is it an improvement that was going to happen anyway?
I have gone back and segmented the data for every team’s season from 1979–2017 (excluding strike years and the year following strike years) to compare teams with new head coaches to teams who kept their head coaches, to see if there is a discernable difference.
This additional data helps to add more context to the win improvements that we see out of first-year head coaches. Generally, they follow closely with the improvements teams experienced who kept their coach. Coming off of 2 to 4 win seasons, you see a little bit of a lift from 1st-year head coaches, and a small gap coming off of 8–9 win seasons compared to teams that kept their coach. But on aggregate, I don’t think you can conclude that a new coach in year 1 has done much more to cause a team to improve their record over what positive regression would have done to their records regardless.
Another way to look at this is the average win-shift in year N+1. Below, I’m showing the average wins in the year following a 2-win season through a 9-win season. I need to mention the caveat that these aren’t large sample sizes. Both groups show similar results, but in most cases, teams that hired a new coach slightly underperformed teams that kept their coach. As an example a 3-win team who keeps their coach will more than double their wins the next year to 6.3 wins, a team who hires a new coach will also improve, but not quite as much (6.0 wins). Without this context, one might attribute a team doubling their win total from 3 to 6 to the new coach, when in fact, it was probably going to happen anyway.
As an example, last year, the Jets had the 4th most Adjusted Games Lost (AGL) due to injuries according to Football Outsiders. That bad luck is not likely to repeat. That bad luck was, in part, why they had a 2–14 record. Of course, Adam Gase needed to go, injuries weren’t the only reason for the 2–14 record, but the injuries certainly helped to pile on the losses and the losses help to fire their coach. Robert Saleh is coming into a situation where positive regression is probable. He will likely lead the Jets to a handful of extra wins. But some extra wins may have happened with or without the new coach as we can see when we start to compare the data for teams with a new coach and teams keeping their coach.
We should judge coaches on many things in their first year (change in culture, play calling, building out the staff, etc) but of course, their W-L record will be judged as well. Perhaps the baseline they are judged against when looking at W-L improvement isn’t the W-L record the season before, but the expected positive regression. For example, the baseline for a newly hired head coach coming off a 4-win season isn’t to do better than 4 wins, it’s to do better than 7 wins.
It’s true that more often than not, new coaches will lead their teams to better records. But the teams hiring new coaches are the teams who were already likely to see the largest increase in wins. This analysis is simply a look at the coaches’ first year after being hired. Hiring and evaluating a new coach is a process that goes well beyond their year 1 record. But I think this is a good example of the correlation vs causation challenge that inevitably comes up when we look for causality between two things.