BFTB Study: New offensive coordinators

A few weeks back, the managing editor here, John Gennaro, asked me the following question: what are the historical results when a team changes offensive coordinators?

Here is my attempt at answering that.

The data

The worst step of doing a study like this, as always, is doing the grunt work to compile the requisite dataset.

Luckily, Pro Football Reference lists each team's coordinators in each team+season page. With a small script to scrape this data for each team and year into a usable, comma-separated file, I was able to take a look at the raw data in Excel and remove data anomalies.

Unfortunately, the data wasn't perfect: there was no uniform way that mid-season coaching changes were accounted for and I found a handful of seasons where a dual HC/Offensive Coordinator was only listed as one. For those of you planning on using data from Pro Football Reference, be aware it'll require some massaging.

While I did my best to fix these by hand, I make no promises that all 414 points in the dataset are absolutely pristine.

The Aggregate Results

The table below holds the aggregate results for various combinations of coaching changes, sorted by the most common form, for all teams from 2001 to 2013.

Total	New HC	New OC	Old OC Became HC	Previous Pts	Next Pts
233	No	No	N/A	363.1	360.6
79	Yes	Yes	No	286.6	305.9
65	No	Yes	No	319.1	338.8
18	No	Yes	Yes	361.2	356.7
13	Yes	No	N/A	385.8	345.0
6	Yes	Yes	Yes	388.1	373.2

Let's go through each scenario other than the Chargers' scenario (No Yes Yes):

No coaching change (No No N/A)

As one would expect, the most common year-to-year result is no change to either the Head Coach or Offensive Coordinator. The aggregate year-to-year difference in this case was -2.5 points. Roughly speaking, natural regression for coaches who weren't fired balances out with natural progression for teams maturing in the same system year-to-year, though the bias towards keeping a coach who is coming off a good season makes the year-to-year difference slightly negative. This isn't profound.

Cleaning House (Yes Yes No)

The next most common year-to-year result is the good old house cleaning. There is one subtly here, in that the previous offensive coordinator in this instance did not become a head coach elsewhere (which actually has happened six times). This is your classic coaching change scenario.

Here, as expected, we see that the previous points scored total is lower than any of the other scenarios. Natural progression is probably expected simply due to the bias that coaches are rarely fired after a lucky season, so there is likely a disproportionate amount of unlucky seasons in the sample. Regression to the mean and whatnot.

The main takeaway here is there's an aggregate improvement of 19.3 points. Not insignificant, but it still brings the team up to an expected points total that is lower than any of the other categories' previous points totals. If there's one thing to learn from this group, it's that Cleaning House can take some time and/or a coaching change may not mask the underlying reason why a team is failing to perform.

Simply Replacing the OC (No Yes No)

Barely less frequent than the 'Cleaning House' scenario was your simple replacement of an Offensive Coordinator. These do not include the Chargers scenario, where the Offensive Coordinator is hired elsewhere as a Head Coach.

As we may expect, the previous points scored total for these teams is between the previous two scenarios: no coaching change and Cleaning House.

It turns out that teams who replace their OC - for reasons other than the OC getting a Head Coach gig elsewhere - improve by 19.7 points, basically the same as the Cleaning House scenario.

At some level, this justifies teams replacing their offensive coordinator.

We need to make a change, but that OC can stay (Yes, No, N/A)

These are understandably rare: just 13 instances in the entire dataset. 2 of these are Tom Moore staying on as offensive coordinator in Indianapolis through coaching changes, while another 2 of them are Sean Payton's suspension and return from suspension.

Other instances include Mike McCoy hanging on through McDaniels to Studesville to Fox and Marvin Lewis hanging on to Bob Bratkowski in Cincinnati.

Because these are rare, the aggregate statistics are greatly influenced by a few anomalies. Generally speaking, though, if you're keeping the OC on board, it's likely you're coming off a good offensive season. Regression is only natural.

What the hell? (Yes Yes Yes)

This scenario surprised me. Since when does a team fire its staff only to see that offensive coordinator sign elsewhere as a head coach?

I should have known better: the Chargers' replacement of Marty and Cam Cameron with Cam's subsequent hiring in Miami is one such example. The Cowher to Tomlin transition, which saw Whisenhunt move to Arizona and a head coaching gig, is another example.

This would also include the rare cases where the OC is promoted to Head Coach: Hue Jackson's in Oakland, for example.

Because of the rarity of these cases, I also wouldn't read into the aggregate numbers other than to say that wow, this appears to only happen to high scoring teams.

The Chargers Scenario (No Yes Yes)

At the highest level, the 2002 Bucs, 2007 Giants, and 2009 Saints won the Super Bowl with a new offensive coordinator, while the 2012 Ravens won the Super Bowl despite switching offensive coordinators mid-season.

More granularly, there are 24 instances of an offensive coordinator leaving one team to become the head coach at another. Interestingly, none of these teams won the Super Bowl. These 24 teams averaged 368 points prior to losing their offensive coordinator, and managed 361 in their first season without him. The other 390 teams averaged 341 points prior and 345 points in the following season.

So what happens once you reduce the dataset to teams who made the playoffs in the previous season, like the Chargers?

There are 30 instances of an offensive coordinator change by a playoff team since 2000. While the average result is a year-to-year decline of 18 points, more than half (16/30) of the playoff teams who hired a new offensive coordinator actually had a positive year-to-year change. In other words, the results are mixed.

However, for only the ones where the head coach stayed on, there are some promising results: only one of the six playoff teams who lost its OC for a head coaching gig failed to return to the playoffs. That was when Mike Shanahan lost Gary Kubiak back in 2006. On the flip side, exactly half of the teams (six of twelve) who failed to make the playoffs in the season prior to losing their OC to a head coaching gig made the playoffs the following year. None of eighteen, however, won the Super Bowl.

On aggregate, however, the 18 'No Yes Yes' teams under the Chargers scenario scored 4.5 points less in the next season. When Mike McCoy left the Broncos in 2013 in a similar fashion to Ken Whisenhunt, the Broncos offensive output actually increased by the largest amount in the entire dataset: a 125 increase from 481 points to 606; we might think of this an an outlier, but there's also a -127 data point in the dataset. The median of the dataset is actually +13.

In other words, there's precedent that the Chargers will be just fine with Frank Reich at offensive coordinator.