An Algorithm and a Super Bowl: How the Cleveland Browns Won It All in 2017

"What's really interesting is sometimes in venture capital and doing start-ups, the whole world can be wrong. No one really executed a game-plan—a team-building architecture—around the 3-pointer." — Golden State Warriors Owner Joe Lacob, The Wall Street Journal

It's March and the sky in Cleveland is low and gray; it's one of those Midwestern late-winter days when it's twilight at dawn and every moment after. Paul DePodesta stood in his corner office at FirstEnergy Stadium, watching the wind whip Lake Erie against the storm walls, a churn of gray on gray. The wind and the waves, he said aloud without taking his eyes from the lake, were the old ways of the football establishment. It was his job to go over the top, to surge past and break through. He was the lake. This was one year ago.

This is what DePodesta saw back in 2016, when he first took over as the Cleveland Browns' chief strategy officer, before the Cleveland Browns were Super Bowl champions, back when they were still the league's laughingstock. "We'd done the data, and you almost don't want to believe it," he recalls. It is always difficult, even for men hired because of their ability to think outside of the box, to look the truth in the face, to stare down something truly new. "But the data was the data, and we had our market inefficiency. This was when we realized that our edge, our arbitrage was that we knew something other teams didn't. We knew we must score more points than our opponent."

Over a year ago, DePodesta excitedly shuffled to Sashi Brown's office, the executive vice-president of football operations. "I remember Paul coming in and exclaiming, 'I got it!'" Brown told me. "And he's laying it out, he's giving me the numbers, which of course are proprietary but which I can tell you were jaw-dropping. He's laying it out and if it wasn't for all the data it might as well have been poetry. It could've been Springsteen, it could've been, like, Optimus Prime."

The data, DePodesta explains now, told him that acquiring productive players, instead of the unproductive ones on which previous administrations had focused, offered an undervalued and little-traveled pathway to what the team called PPD—that's positive point differential, which DePodesta's data team, The Brown Squad, discovered was also strongly correlated with victory totals. The problem for the Browns is that they had negative point differential players—or, as DePodesta called them, Below Average Developers, or BADs. The more BAD players you have, the less likely your team is to win. "I wasn't sure I wanted to believe it, either," DePodesta said. "But the data backed me up."

But this wasn't the end—it was the beginning. DePodesta went down to the sub-basement and gathered The Brown Squad, a team of top analysts recruited from the best schools in the country, and told them that they had another task. "He goes, 'This is what you're going to be working on for…well, until you get it right,'" one Brown Squad member recalled. The task was to develop an algorithm that could tell them which players were BAD and which ones were not. And then DePodesta told them the fridge was stocked with soylent, turned around, and left. The door locked behind him, from the outside.

"The first thing we learned," said Kevin Meers, a football research analyst who graduated cum laude from Harvard University, "was that our quarterbacks were BAD."

After weeks of intense data analysis, Meers finally found the key element, the one every other team was ignoring. "It's amazing that no team had emphasized touchdowns before, but our data-dive suggested they were a market inefficiency." In a league that emphasizes field goals and points-after-touchdowns, the Browns had discovered a property that paid off twice or even six times the rate of the competition's targets. "We could go 6x," DePodesta said, "with this one weird trick." And no one else was there yet.

Shortly after the algorithm came out of beta, they ran it against the entire NFL player pool, including NFL Draft prospects. "Our players were all BAD," DePodesta shrugged. "Every last one of them. BAD."

The Browns were coy about what happened next—trade secrets, you understand. They need to protect their intellectual property, which is why the algorithm they developed resides on a password-protected server that requires two-factor authentication to access. "But," Brown told me with a coy smile, "we put a real emphasis on signing not BAD (nBAD) players, and using those players to score more points than our opponents."

What the Browns did next was DePodesta's true stroke of genius. By signing Robert Griffin III, DePodesta took advantage of the greatest market inefficiency in the game: the Washington Redskins. By their research, a large number of former nBAD players became BAD when they joined Washington, and vice versa. By signing a former Washington quarterback, DePodesta surmised, they could get a player who was no longer BAD on the cheap.

It came as no surprise to the Browns when RG3 took the league by storm, throwing for 30 touchdown passes, running for 10 more, and leading the team to hoist the Lombardi Trophy. "By scoring more points than our opponents with nBAD players," Meers recalled, "I really think we engineered a winning strategy."

The problem for the Browns is that getting to an innovation first doesn't come with squatter's rights. Winning strategies are adopted by others, market inefficiencies are suddenly the province of late-arriving copycats, and the first-mover advantage disappears. "Sure, we're aware of that history, whether it's in finance, Pharma, other sports, or even old-school innovations in manufacturing or the printing press," DePodesta told me in his office, his Cleveland Browns Super Bowl Champs hat bumped sideways. "But we're confident no one can replicate what we've done. What the Cleveland Browns have accomplished this year, this is once in a lifetime."