Computer simulations helped Cincinnati choose new head coach Luke Fickell

The template script for an athletic director explaining his or her hire of a football coach is pretty clear, and ADs are good at sticking to it.

“Lots of exciting candidates ... did/didn’t use a search firm ... attracted to the specific coach because of his enthusiasm, experience, recruiting/tactical ability/whatever.”

Every one of these ADs is a different person and likely follows a slightly different process, but what we usually hear isn’t all that different.

Call Cincinnati an exception, then. Earlier this week, GoBEARCATS.com published an inside look at athletic director Mike Bohn’s process for replacing Tommy Tuberville. Eventually he chose longtime Ohio State assistant Luke Fickell, but he got help from at least one nerd in the process.

But long before Bohn interviewed Fickell, he and his staff had done extensive research on Fickell, including the use of data analytics to vet all 12 of the candidates on Bohn’s original list. Bohn said he had never used such research before during a coaching search, but he decided to tap into the expertise of a UC graduate who had a past relationship with the athletic department and expressed a desire to help in the search. [...]

According to Brandon Sosna, Bohn’s chief of staff, a simulation on each candidate starting with the personnel on the current UC roster was run 100,000 times to produce a four-year projection of how the team would perform.

OK, color me intrigued.

“There are a few characteristics of coaches that are consistent no matter where they go or what they do,” Sosna said. “Things like turnover margin tend to follow coaches. Penalties per game tends to follow coaches, and tempo. You’re either an up-tempo coach or you slow down but you don’t tend to change over time.”

As SB Nation’s resident football stat nerd, I feel I have let myself down by never really looking at data in this way. But it’s an intriguing thought. And it would make sense that things like turnovers, penalties, and tempo would follow a coach from one school to another.

All of those things are, to some degree, conscious choices. Your defensive philosophy dictates whether you are playing things conservatively or going after the ball. Your team’s committed penalties aren’t necessarily tied to wins and losses, but they reflect how closely to the legal-or-illegal line your players tend to play. And your tempo is obviously a conscious choice.

I’m not completely sure how you use personality stats to project quality, especially not without knowing what assumptions are being made about Fickell’s recruiting ability. Still, to take a look at the strengths and weaknesses of the current roster, project some level of recruiting value, and apply it to personality stats is impressive and, for a rather staid field, exciting.

The computer analysis projected UC’s win-loss record, points for and against, and where the Bearcats would finish in conference play based on what is currently known about UC’s schedules during the next four years. Sosna said Fickell gave the Bearcats their best chance over a four-year period to win a conference championship at about an 80 percent probability. [...]

[H]e also liked the fact that Fickell was a defensive coach. ... “The performance of the Ohio State defense was the same,” Sosna said. “They were still a top-10, top-15 defense every single year. The fact that he was able to continue to scheme such a great defense despite the polar opposites in terms of coaching styles on the offensive side, his ability to keep that continuity of performance on defense was really valuable.

Again, recruiting assumptions are a tricky thing, and I’m not sure how accurate you can expect to be when a coach is leaving a Power 5 conference. And as a whole, this entire process could be doomed if the data being analyzed (and the people doing the analysis) are faulty.

But using models to at least create an initial hierarchy of candidates is cool. And if Bohn were to use this information as a jumping-off point to talk to connected sources across the country — including people who would know the UC job pretty well — that’s how data should be used.

Armed with that information, Bohn talked to sources across the country, including former UC head coach Mark Dantonio, now the head coach at Michigan State, and OSU’s Urban Meyer, to learn more about Fickell.

Well then.

It is a common assumption that stats are intended to answer questions, when really, they’re intended to create the best possible questions to ask. And if Fickell succeeds at Cincinnati — conference title or no conference title — hopefully it will encourage other ADs to follow a similar path.