The Accountant

Banker by day, assistant football coach by afternoon, Jeremy Fisher first built his complex analytics database for this small Northern Kentucky program nearly two decades ago. And that intense dedication to data has had three-time defending state champ Beechwood punching above its weight class ever since.

A vis­it to Pro Football Focus’ pala­tial Cincinnati mega-lab two years ago did won­ders for Beechwood High School’s staff, includ­ing its intense­ly-dri­ven ana­lyt­ics guru Jeremy Fisher. But odd­ly enough, it was more for what it con­firmed rather than revealed.

Over the course of a day, this small school out of sub­ur­ban Fort Mitchell, Kentucky was shown vir­tu­al­ly every step of how the mag­ic is made. The revered foot­ball ana­lyt­ics data-min­er is major­i­ty-owned by Sunday Night Football ana­lyst Cris Collinsworth and counts more than 60 major Division I foot­ball pro­grams and all 32 NFL teams among its clien­tele. Fisher and his fel­low coach­es peri­od­i­cal­ly found them­selves chuck­ling amongst them­selves through­out the demon­stra­tions — when asked what was so fun­ny by PFF staffers, they replied, Oh noth­ing, we’ll show you at the end.”

The end came about. Now it was Fisher’s turn to take the saber­met­ric wiz­ards through his work­flow process. Project man­ag­er for Citibank by day, foot­ball coach by after­noon, Fisher has meld­ed those two dis­ci­plines into a remark­ably effi­cient data­base that has kept dozens upon dozens of data columns on every sin­gle one of the Tigers’ plays for near­ly two decades. And the data-hunt­ing process — the ten­den­cies, the acute atten­tion to details on for­ma­tions and posi­tion­ing — near­ly mir­rored PFF’s own.

Image Credit: Courtesy of Beechwood Photos

The first thing the PFF staffer remarked? You guys don’t real­ly need us, do you?”

The sec­ond thing was, Do you want a job?”

And, well, that’s just the thing. Fisher, who spe­cial­izes in Management Information Systems and data­base designs for one of the world’s largest banks, loves his day job. Loves the num­bers. Loves what he can do with the num­bers. Loves the lim­it­less pos­si­bil­i­ties they hold. 

And play­ing with num­bers all day has only inflamed his pas­sion for foot­ball ana­lyt­ics over the years.

The mar­riage of these two prin­ci­ples goes all the way back to 2002, when a coach­ing change ele­vat­ed then-defen­sive coor­di­na­tor Noël Rash into the head coach role. Rash came from a vast­ly dif­fer­ent back­ground than his pre­de­ces­sor — an assis­tant at one of the most sto­ried high school pro­grams in Ohio state his­to­ry, Archbishop Moeller — and want­ed to look at things in a new way. 

He want­ed to take a deep­er dive into the data, and then he want­ed to look at it from a dif­fer­ent per­spec­tive. To Fisher, Rash laid out all of his thoughts one day, and then said the mag­ic words, Can you make this work?”

Goodness gra­cious, did he ever.

Where’s the game going? Our part­ners at X&O Labs show you the lat­est trends, and how you can adapt.

Download the Report

Years before Hudl had even got­ten its foot­ing in the mar­ket, the Beechwood staff was already keep­ing any­where from 20 to 44 pieces of data on every sin­gle play, export­ing the infor­ma­tion into a Microsoft Access data­base that host­ed all spe­cif­ic play data obtained in break­downs going back to 2006. From there, Rash got a report­ing plat­form with all the infor­ma­tion in the exact for­mat he want­ed. And with Fisher’s back­ground in a pletho­ra of report­ing soft­ware, Rash could get this infor­ma­tion how­ev­er he wants, with what­ev­er fan­cy visu­als he tastes.

Fisher was then (and still is) chal­lenged by his supe­ri­ors to go a lit­tle fur­ther down the rab­bit hole. That means going beyond the box score to chart out a num­ber of addi­tion­al items rang­ing from the sim­ple (field vs. bound­ary, strong/​weak run­ning by for­ma­tion and down) to the com­plex (chart­ing indi­vid­ual play­er touch­es by posi­tion, for­ma­tion and tar­gets). Fisher even goes as far as to use a stop­watch app on his com­put­er to time how long it takes the quar­ter­back to release the ball once it’s snapped. The Tigers adjust their pass cov­er­age based on the length and tim­ing he measures.

By any stan­dard in the high school stratos­phere, what the Tigers have in place is aston­ish­ing. Thanks to an acute atten­tion to ten­den­cies and tells, they often find they know what to expect when oppo­nents break the hud­dle. In 2016, their first of three straight state cham­pi­onship-win­ning sea­sons, Fisher esti­mates that his defen­sive staff accu­rate­ly guessed their opponent’s exact play near­ly 63 per­cent of the time.

The next sea­son, that num­ber dipped to 48, and then last year it was 52 as they made it a Class 1A three-peat.

Image Credit: Courtesy of Beechwood Photos

Just how valu­able is that infor­ma­tion? Consider the Tigers’ final two oppo­nents of the 2018 season. 

In the Class 1A state semi­fi­nals, their touch-track­ing data, which Fisher typ­i­cal­ly com­part­men­tal­izes into per­son­al touch­es, posi­tion touch­es and for­ma­tion touch­es (includ­ing tar­gets into those fig­ures), con­clud­ed that Campbellsville’s star ath­lete touched the ball on 80 per­cent of the offense’s snaps. By the end of the first half against Beechwood, that star ath­lete had totaled minus-24 yards from scrim­mage, as the Tigers went on to win in a romp.

The next week, in the 1A state final on the University of Kentucky cam­pus, Fisher’s data found that Pikeville’s star quar­ter­back was not only an excep­tion­al run­ner who got the ball 70 per­cent of the time, but was also used aggres­sive­ly to com­mand pace ear­ly. Pikeville’s aver­age game-open­ing dri­ve last­ed only six plays, with their star get­ting five touch­es. But against the Tigers, their first series end­ed in a three-and-out — and no touch­es for their pri­ma­ry play­mak­er. Pikeville’s QB also nev­er kept the ball on third down, which influ­enced Beechwood to show com­plete­ly dif­fer­ent pack­ages on first and sec­ond downs. All of it added up to a nail-bit­ing 21 – 20 win.

To hard­en itself for the state play­offs, the Tigers will rou­tine­ly sched­ule oppo­nents dur­ing the reg­u­lar sea­son that are three or four class­es above them. And while in some years they may have lux­u­ries in size or ath­leti­cism, those are most­ly excep­tions. Often, they are under­sized against oppo­nents. But this deep well of infor­ma­tion allows them play fast, with­out hav­ing to think too much out there.

We don’t want to come across like we feel we’re bet­ter than any­body,” Fisher said. But at the same time, we also don’t want peo­ple to real­ize how much infor­ma­tion we real­ly know and under­stand. There are very few games that we go into and say, ugh, we don’t have a chance.”

Fisher esti­mates that the Tigers’ data­base has data on more than 50,000 indi­vid­ual plays going all the way back to 2002. As the game con­tin­ues to evolve in so many abstract ways, and as your typ­i­cal high school offense becomes less and less pre­dictable in this age of mega-data, the Tigers will con­tin­ue to tweak and enhance this elab­o­rate workflow. 

I hate to come across as though I have some sort of spe­cif­ic knowl­edge,” Fisher said. But I’m not gonna lie, I’m fair­ly proud with myself of the fact that I’ve been able to find some suc­cess relat­ing between the two jobs that I have, to take what I do at Citibank and be able to apply data analy­sis, to be able to inter­pret results and infor­ma­tion, and be able to apply it to football.”