Even the founders of AccuScore, a tech startup that runs statistical simulations of games to forecast scores, concede it’s a nerdy way to follow sports.
The L.A. company has developed software that mathematically calculates everything from player performance and weather to coaching strategies of teams playing in professional and college sports leagues, including the National Basketball Association, the National Football League, Major League Baseball, and NCAA football and basketball.
It uses patent-pending algorithms to simulate a single game, one play at a time. The same game is then simulated 10,000 times, distilling the possible configurations of player statistics and coaching strategies to get the most accurate forecast of the outcome.
The AccuScore analysis of its forecast, such as whether a team can increase its chances of winning by using a certain strategy, becomes fodder for news media, from mainstream broadcasters and publishers to fantasy sports enthusiasts.
AccuScore landed one of its first contracts in 2005 when Yahoo Sports asked the company to provide statistics on past games and then compare them with Yahoo’s data. The company outmatched Yahoo’s forecasts.
After sealing its first licensing agreement with Yahoo, the company, now with seven employees, locked in contracts with ESPN and CBS Sports.
“Sports is often about debate, or ‘My team is better than your team,'” said Patrick Stiegman, vice president and executive editor of ESPN.com. “This is another tool to allow fans to know a little bit more about their teams than their best friends or win their arguments around the water cooler.”
As a precursor for the NCAA basketball season, ESPN.com created an online bracket where users can simulate every combination of matchups based on AccuScore data.
It’s content Stiegman calls “sticky” or the kind that attracts and keeps users on the site, the kind that advertisers love.
Go Bears
A turning point for AccuScore, which is bootstrapped and shares its office space with another startup, was in December 2006.
With three weeks left in the regular season, AccuScore predicted that the Chicago Bears would host and beat the New Orleans Saints in the National Football Conference Championship Game. ESPN.com listed predictions by AccuScore and eight football experts. Only AccuScore forecast a Bears’ victory.
The Chicago press noted the accurate pick and now AccuScore data is regularly used by sports writers across the country, including those at the Los Angeles Times, for pregame analysis.
One challenge for the company has been turning 500,000 rows of data on its turbocharged Excel spreadsheet into a compelling story for the audience.
“This is something no one else has in the market, so we have had to spoon-feed the audience at times, simplifying the analysis as much as possible,” said Gibby McCaleb, the company’s chief operating officer.
The information is so advanced that media outlets that license it have confided to AccuScore executives that they’re happy to defer to their brand to market the data.
“The overarching challenge is to figure out the long term use of this kind of data,” said Stephen Oh, the company’s chief product officer. “We frankly haven’t figured out how best to use it.”
Experts are impressed with the power of the software.
“I haven’t seen anything else like this,” said Scott Swanay, president of Fantasy Baseball Sherpa, a sports forecasting Web site. “They take into account injury and weather, which are the most important external variables.”
Applying more external variables, such as how a player’s newborn baby or a death in the family would affect his performance, would make the projection even more advanced, Swanay said.
‘Time machine’
The origins of AccuScore trace back to Oh’s days as a Ph.D. candidate in an anthropology lab a decade ago, working on mathematical simulations at University of Michigan. He felt he’d have more fun using the same models for sports.
“Studying human evolution, you could never figure out which theory was right even with 95 percent confidence, because there’s no such thing as a time machine,” Oh said. “When it comes to sports, even a 55 percent chance of winning gets people really excited.”
For example, when AccuScore reports that if the New England Patriots pass the ball more, it would increase the team’s chance of winning by 5 percentage points, that data might get in an ESPN headline. However, such mathematical calculations would carry little weight in the scientific world, Oh said.
Oh left the scientific world, abbreviating his studies by completing a master’s program, and got a job at an L.A. marketing agency. In 2004, McCaleb and Jason Manasse, now chief executive of AccuScore, made him a sales pitch. McCaleb and Manasse offered to revamp the marketing agency’s Web site.
But Oh had another idea.
Out of his love for sports, he had built a mathematical model for simulating games and asked whether the two men could market that instead.
The three founders pooled $400 to incorporate the company and subsequently raised $100,000 in seed capital.
Shortly after, the founders drove up to Sunnyvale, Calif., to show their data to Yahoo. “What we realized in that meeting was that we had figured out things they hadn’t yet,” Manasse said. “That was the beginning.”
