Nik Bonaddio was in the hot seat on Who Wants To Be A Millionaire one night back in 2009. He had coasted through the first few questions. With $100,000 in the bank, Bonaddio was stumped by a question that asked him to identify which '80s movie featured a character named Axelrod. If you watch the tape of his appearance , you'll see that Bonaddio appears to make a quick and seemingly painless decision to walk away with the $100,000, even though he had narrowed the choices down to just two: Endless Love and An Officer and a Gentleman. Even with a 50% chance of being right, he knew that walking away with $100k was the right thing for him to do.
Sitting in a Lower East Side coffee shop three years later, Bonaddio, 30, CEO and Founder of numberFire, explained the strategy: "There's a concept in statistics called expected value. On that question I had it down to two choices. I was at $100,000, I was going for $250,000, so the expected value was $125,000. So the statistically rational decision would have been for me to guess. But I just couldn't risk it."
He couldn't risk it because he was not an average game show contestant. Bonaddio had come on the show for a very specific purpose—to raise cash for his startup company. At the time, Bonaddio was working as the creative director for an app development company called SNAP Interactive, but the company Bonaddio wanted to start had been percolating in his mind since he was a boy in Pittsburgh. "I used to watch the Steelers every week," he recalls. "I noticed that the Steelers would always run the exact same play on fourth down. I must've seen the same play 50 times. I turned to my dad and said 'I know what's coming, why doesn't the team know what's coming?' The data is there. Someone is collecting the data for every single play in the NFL."
Long before the world knew about the concept of "moneyball," the idea for numberFire had occurred to Bonaddio. The idea was to take all the data that exists for major league sports and put it into a model that could accurately predict the outcome of any given matchup, based on historical data from similar matchups. "Twenty years ago people thought that stocks just moved. Then they developed algorithms that allowed hedge funds to be wildly successful. For a long time, people have thought the same way about sports: that it 'just happens.' But sports are all about numbers. The stats, the scores, they are all numbers. There are reasons for everything, and they are understandable if you look at the data," Bonaddio pauses then adds excitedly, "I'm very lucky to live in a time where there's more data than ever before."
Bonaddio officially launched numberFire in September 2010. A few months before his official launch, he headed to TechCrunch Disrupt to premiere the site. He came to the conference with a one-page demo, but was surprised when he arrived to learn that the conference organizers expected him to have a full working prototype. Despite the miscommunication, at the conference David Tisch became a fan, and urged Bonaddio to push forward. Tisch is now an investor in the company. NumberFire's now COO Sean Weinstock dropped out of NYU Law School to join the company, and Keith Goldner, who is the company's "Chief Statistician," turned down a job with the Philadelphia Eagles to be part of the numberFire team. Today, their office in New York is a single table in a shared workspace with several other startups.
Bonaddio spent several months inputting sports data that went back decades in order to build a predictive algorithm. The numberFire algorithm takes into account all the same individual players' game statistics (like yards gained passing for a quarterback), which many standard statistical models account for. But numberFire's algorithm gives those individual statistics richer meaning. They put those statistics in context by adding proprietary statistics (like their own measure of "effective yards gained passing"), comparable historical players and situations, the strengths and weaknesses of the player's team overall, and the strengths and weaknesses of the opposing team. The prediction is based on a union of these data sets. This means that numberFire's algorithm will suggest to a fantasy football participant which NFL player they should start in any given game and who is most likely to win the game.
Or as the company eventually put it, "We do all the number crunching—all you have to do is win."
An estimated 32 million people play fantasy sports and the total estimated fantasy sports economy is between $3 and $4 billion each year. With so many people so invested in their fantasy leagues, numberFire serves as a much refined recommendation engine for fantasy players to fill out their play cards. NumberFire also has a strong record of accuracy. It beats the Vegas NFL odds 58% of the time. In fantasy sports overall, numberFire beats their main competitors 61% of the time. Late last year, the website Fantasypros.com hailed numberFire as the most successful prognosticator of week 15 of the NFL season.
If numberFire can continue to enhance its growing reputation for accuracy, the applications will go way beyond fantasy players. Sports teams, leagues, coaches, scouts, and athletes are interested. So too are professional gamblers. But Bonaddio is clear, "numberFire is not a gambling company. We're a data company that is agnostic to what anyone does with the data."
Currently numberFire offers projections for teams in MLB, the NFL, and the NBA. This week, they are entering the NCAA March Madness fray for the first time, with a Facebook app that will provide full bracket recommendations to all users. They've partnered with Thrillist and StubHub to drive traffic to the app and expand their current user base of approximately 25,000 users. They've spent zero dollars on marketing, but since their launch, traffic has grown monthly by 100%.
Much of the fantasy sports play occurs on ESPN and Yahoo Sports' platforms. They both use a company called AccuScore, which has long dominated the field of predictive sports analytics. Bonaddio believes that AccuScore has pioneered the market for paid sports statistics—and that, without AccuScore, numberFire would not even exist. But that doesn't stop him from hoping some day to replace AccuScore as the premier data source for ESPN and others. Already, numberFire's projections have appeared on the websites of Sports Illustrated and ESPN. Bonaddio believes that statistics can be a valuable tool for sportswriters, but that they enhance sports commentary without replacing it, "The data can help writers make stronger points and qualify their assertions. But it's always going to be hard to use data to predict rookies or players who come out of nowhere. We're always going to need some kind of qualitative analysis."
Early on, naysayers told Bonaddio he couldn't make money with sports data, but he says that last year's film Moneyball helped some people understand what he was doing and why it matters—at least conceptually. Baseball, seen by Bonaddio as a sport that has been historically more accepting of "nerdiness," has always had at least a small coterie of managers, fans, and sportswriters fascinated by statistics. But what Michael Lewis demonstrated in writing the book on which Moneyball was based, and what the film popularized, was the truth that statistical analysis, if done correctly, can be a valuable tool in identifying the right players for a team, creating winning teams, and ultimately, in making money and succeeding in the business of sports.
NumberFire closed a seed financing round of $750,000 in January that reportedly included New York-based venture firm RRE and several angel investors. Bonaddio has big plans for the site's growth. He wants to expand it to cover more sports and particularly more global sports. Much of the market and attention in fantasy sports is focused on football, but Bonaddio says that hardly any statistical analysis exists for golf or tennis, let alone more internationally based sports like cricket and soccer.
NumberFire's use of data is designed to enhance the enjoyment and experience of sports, not to reduce it to "mere" statistics. Bonaddio knows that sports will never achieve 100% predictability and will always be highly dependent on the human factor. Bonaddio, Goldner, and Weinstock are all big sports fans themselves. They love to get together and watch a game, although these days that often means working while watching. So it begs the question, Is there a danger in trying to boil down sports to pure statistics? "It can't be all feeling or all data. But we can't predict the intangible," says Bonaddio. "In this year's NFL season, the Giants weren't the best team in most of those games. But they were trending in the right direction. They won the Super Bowl, but they shouldn't have won according to the data. Being 'the team of destiny,' getting along together on and off the field, those are all intangible, you can't turn those into data points."
David D. Burstein is a young entrepreneur, having completed his first documentary 18 in '08. He is also the founder & executive director of the youth voter engagement not for profit Generation18. His book about the millennial generation will be published by Beacon Press in early 2013.
[Image: Flickr user Tyello]