Football Coaching Personality: The Mad Scientist

The rea­son usu­al­ly starts with the tac­tic he’s most famous for: His strat­e­gy of nev­er punt­ing, and almost always attempt­ing an onside kick after a score. And we’re not talk­ing about just a sim­ple squib-it-and-pray approach either. No, Kelley’s arse­nal of onside kicks spans the entire spec­trum, from the stan­dard to the absurd. Baltimore chops. Middle bunts. Drop kicks. Tee-less ​“copter” kicks, lay­ing the ball flat on the ground and boot­ing a nose for a knuck­le­ball effect. Soccer-style free kicks involv­ing two, some­times three, kick­ers feign­ing a boot before one of them lobs it towards the sideline.

The vari­ety at times seems lim­it­less, with fun names like ​“NBA” or ​“Jelly” or ​“MLB” to boot. And in an age where the nation’s best high school place­kick­ers are clear­ing the cross­bar from 60 yards out, Kelley’s tac­tic pri­or­i­tizes close-range accu­ra­cy over leg pow­er, no mat­ter how Herculean. Perhaps that’s why, unsat­is­fied with his crop of kick­ers at the begin­ning of the 2018 sea­son, Kelley went out and recruit­ed one from Pulaski’s girls’ soc­cer team. She flour­ished, manip­u­lat­ing the ball with a sur­gi­cal precision.

Inevitably, the con­ver­sa­tions then veer into his oth­er schools of thought that are all rein­forced by hard data. Like what fuels his ongo­ing exper­i­men­ta­tion with rug­by-style lat­er­als in his pass­ing plays. Or why he rails against the fade route, despite its pop­u­lar­i­ty in the NFL. Or how he cal­i­brates an avant-garde offen­sive play­book that nev­er ceas­es to pro­duce video-game num­bers, yet still laments that after a 94-catch sea­son from his All-American tight end last fall, ​“I prob­a­bly used him as a decoy too much.”

Sensing a pat­tern? Kelley has banked his whole career on going against the grain, on nev­er accept­ing con­ven­tion at face val­ue, and along the way has rewrit­ten the book on prob­a­bil­i­ty. There’s plen­ty of detrac­tors out there, but they can’t argue with his suc­cess — a 191 – 27-1 record and sev­en state titles since he took over Pulaski in 2003, includ­ing five titles this decade.

Friday nights at the Kelley house­hold includes catered food for his coach­ing staff, most of who will stay until the crack of dawn, lap­tops in tow, extract­ing data and curat­ing scout­ing reports. He then con­tin­ues the X-and-O’s med­ley the next day at what he calls ​“Buffalo Wild Wings on steroids, but with local food.” But when it comes to the off­sea­son eval­u­a­tion, he prefers to oper­ate alone.

“Golly, I’m just so hands-on-y, it’s like a sci­ence exper­i­ment — you don’t want to taint it. You know what I’m say­ing?” Kelley laughs.

Kelley, the 2016 USA Today National High School Coach of the Year, first began toy­ing with his punt-less strat­e­gy when he took over as Pulaski’s head coach in 2003. But he went all in on no punts in 2007, when the Bruins reached the Class 5A quar­ter­fi­nals. Over the last 12 sea­sons, they have only punt­ed eight times total, with half of those 12 sea­sons fea­tur­ing no punts at all. Their last punt of any kind was in 2017.

When it comes to kick­offs, Kelley gen­er­al­ly abides by his own ​“Rule of 21”. If the Bruins are lead­ing or trail­ing by three scores or less, oppo­nents can prob­a­bly expect an onside kick of ency­clo­pe­dic diver­si­ty, rang­ing from the stan­dard to the bizarre.

Like so many of his oth­er unortho­dox deci­sions, this wild­ly unique pathos of no punts and aggres­sive­ly kick­ing onsides derives from sta­tis­ti­cal analy­sis that sug­gest­ed foot­ball coach­es gen­er­al­ly under­val­ue pos­ses­sion. Kelley also sees this as an oppor­tu­ni­ty to fine-tune oth­er parts of their game — in oth­er words, those 20 min­utes spent every week on punts could be re-allo­cat­ed to 20 min­utes spent on an offen­sive play that makes all the dif­fer­ence on Friday night.

“What I call it is applied ana­lyt­ics. That’s where I think we [as a whole] are strug­gling a bit,” Kelley said. ​“They can look at the num­ber and go, ​‘you should or shouldn’t in this moment.’ But they’re not look­ing for what I call the ​‘but­ter­fly effect’ of how it makes every­thing else you do in prac­tice and dur­ing the week lead­ing up to it, and that kin­da stuff.”

Kelley relates his curios­i­ty to a word search puz­zle (and, per­haps coin­ci­den­tal­ly, he still prefers pen-and-paper over spread­sheets). When he peels back the ana­lyt­ics to eval­u­ate some­thing, he’s look­ing for two or three things in par­tic­u­lar. But then, five or six oth­er ideas come to mind dur­ing the process, and so he starts eval­u­at­ing those too, each side of the brain per­pet­u­al­ly feed­ing the other.

And some­times, what he uncov­ers defies his own con­ven­tions. Evaluating last year’s offense, for instance, a pass­ing play he didn’t think much of because it nev­er scored was actu­al­ly 16-of-17 on the year for 210 yards. Meanwhile, a run­ning play he thought was a home run because of the two long scores it pro­duced, was actu­al­ly his worst run­ning play — out­side of those two gains, it most­ly went for neg­a­tive or no gain. Now that run is out of the 2019 play­book. The human con­di­tion, Kelley says, caus­es us to remem­ber the explo­sive moments over the steady ones, cloud­ing out eval­u­a­tions as a result. That’s why he feels he needs to go so deep with his data.

What all that research leads to is an adren­a­line rush of an offense, one that is often among the nation’s high­est-scor­ing. Over the last five sea­sons, the Bruins have aver­aged 50.3 points, 563.8 yards, and an absurd 8.2 points per play. This break­neck style stems from a belief that even one big-yardage play has a dra­mat­ic effect on scor­ing chances, a hypoth­e­sis that was con­firmed once again this off­sea­son when he real­ized near­ly 88 per­cent of his dri­ves in 2018 that fea­tured a play of 20-plus yards end­ed in a touch­down. Over that same peri­od, Kelley also found that his team was five times less like­ly to score on a dri­ve if it com­mit­ted even one penal­ty. That means three yards and a cloud of dust is far from ideal.

So what is ide­al? A fast-paced attack pred­i­cat­ed on spac­ing and built around plays with the high­est like­li­hood of going for 20 yards. That means more rhythm pass­es that max­i­mize receivers’ chances to pile up yards after the catch. Pulaski cap­i­tal­izes on inter­me­di­ate throws between the hash­es, deploy­ing mov­ing tar­gets into the space between the line­back­ers and safeties. Not only are they a high­er com­ple­tion per­cent­age, they’re also an eas­i­er throw for quar­ter­backs to make, as opposed to ones out­side the num­bers towards the side­line. Those low­er-per­cent­age throws require place­ment into a tight win­dow that even the best high school QBs can strug­gle with, and often end with the receiv­er walk­ing out of bounds after the catch.

Kelley also uses a lot of easy-to-install plays that his research has found to be prac­ti­cal­ly unstop­pable. Rubs and mesh­es. But also throw­back screens. Second and third-lev­el RPOs. And of course, one of the most unstop­pable run­ning plays of all-time — the counter trey.

Curiosity, and the con­trar­i­an mind­set that comes with it, led Kelley to an even more rad­i­cal dis­cov­ery in recent years. Using num­bers from ESPN.com’s col­lege foot­ball data­base, he found that when two play­ers touched the ball on a play, their chances of going for 20 yards hov­ered around 10 per­cent. But when three play­ers touched the ball, that like­li­hood dou­bled.

Of all things, the most sure­fire way to this goal was actually…a hook-and-ladder?

This dis­cov­ery led to Kelley’s most rad­i­cal idea to date: installing rug­by-style lat­er­als on the back end of pass­ing plays, a twist on one of football’s most cel­e­brat­ed trick plays. When the play calls for it, down­field receivers will forego block­ing and instead trail the ball-catch­er, yelling out when there’s an oppor­tu­ni­ty to lat­er­al the ball backwards.

In 2015, his first sea­son with the exper­i­ment, Pulaski con­vert­ed 30 suc­cess­ful lat­er­als. Last sea­son, with All-American tight end Hudson Henry’s brutish 6-foot-5 frame box­ing out, the Bruins con­vert­ed eight, with the sure-hand­ed future Arkansas Razorback hav­ing a hand in all of them.

So where does this curios­i­ty come from? Is it born out of neces­si­ty? Does it stem from a desire to stay one step ahead of the competition?

Kelley says it comes from the will to win — and, sub­se­quent­ly, the will­ing­ness to do what­ev­er it takes to win.

“People like to say the ques­tion is do you want it more — no, every­body wants it the same,” he says. ​“But did every­body want it the same when they were prac­tic­ing and com­ing up with what they were going to do in the game? Did they want it enough to try some­thing dif­fer­ent, not know­ing the out­come — not punt­ing, onside kicks, pitch­ing the ball around? To me the big dif­fer­ence is tru­ly will­ing to win at the risk of sac­ri­fic­ing your ego, at the risk of peo­ple boo­ing you, deal­ing with social media, deal­ing with the media in a neg­a­tive way.

“So I do think it comes down to my will­ing­ness, I’m will­ing to do all these things that I real­ly think give us a bet­ter chance to win, but at the risk of sac­ri­fic­ing all those neg­a­tives, along with, if it fails mis­er­ably. A lot of peo­ple just aren’t.”

Never before in the his­to­ry of our game has so much advanced data been so read­i­ly avail­able at even the most grass­roots lev­els. And it’s not going away any time soon. If any­thing, we’ve just begun to scratch the sur­face, a point Kelley ham­mers when asked about what’s under­val­ued in today’s intense cli­mate of analytics.

“I think that as we go along, it’s still real­ly new in the his­to­ry of things. Data analy­sis is still real­ly young, and I think we’re going to explore bet­ter ways to use it, and there’s a mil­lion things out there that we’re just get­ting start­ed on that we can do,” he says. ​“And so I think the under­val­ue is, in the end, we think we know too much, and when you do, that slows down growth. It slows down par­a­digm shifts. It slows down everything.”