StatSheet Network Automates Hundreds of Sports Stories from Databases
The buzzers sound and Robbie Allen's computers start humming. Within hours they have written and posted stories about all the college basketball games played that day.
The content, on 345 teams, is essentially untouched by human hands. It has been generated from massive databases of statistics, sample paragraphs and phrases and stat feeds from the games.
The StatSheet Network, which launched this month, provides game advances, game stories, game notes, historic look-backs and interactive graphics on game flow, player comparisons and game stats. Each game gets stories created for the perspective of both team and their leagues. The network made season predictions, it forecasts game outcomes and it generates statistical reports on players of the week -- and all other players. It can choose freshmen or sixth men of the week. StatSheet has a mobile version. It tweets scores during the games, invites reader comments, is customizable and available on mobile devices.
Allen has automated sports writing.
He does not plan to stop with college basketball. He believes the system, once perfected, can be used for baseball, football and then hockey and other sports. "We're going to be supporting every major sport within the next year," he said. Beyond sports, he said in a phone interview, he sees possibilities for daily market reporting, which is highly statistical.
But for now this University of North Carolina basketball fan will stick with his passion. Allen has written books and magazine articles and started a blog about his favorite team in 2003. Keeping up with one team that way was overwhelming and he thought about hiring others to help him run such a site, but was leery of signing writers and negotiating contracts.
An information technology professional, he decided to go with fewer employees -- seven of his nine are developers and just one is a writer -- and to cover all the teams.
Allen said StatSheet received $1.3 million in funding from Valhalla Partners in August. Long-term, his business model is a familiar one, with advertisers, sponsors and stores that sell tickets and merchandise.
There still are glitches to be worked out. While the statistics and their interactivity are impressive, the system is not set up to capture quotes, and photos would help. Allen said he is negotiating for photos. Team logos would help, too. The writing is sometimes a little off, but most of the stories I read on my favorite team's site, Spartan Ball, were clear.
"The content may be a little rough in some places, but you don't have many Pulitzer winners writing sports, so the bar isn't very high," Allen said. "Someone sent me an article written by a well-known sports columnist, and our algorithm could have generated that."
Allen, who said his favorite human sportswriter is ESPN Sports Guy Bill Simmons, said blogging Carolina basketball "was a bit of a chore."
"Writing sports content is still really tedious," he said. "You're covering a performance, stats and players. There is no way to integrate box scores into a blog legitimately."
This led him to StatSheet. He said, "I'm not out to put anybody of a job and I don't think I will. If anything, it's going to free up writers to focus on the more interesting events of sports instead of having to sit down and do a bunch of number crunching."
I originally wrote about Allen in a post about robotic journalism and told him people had talked about having computers generate obituaries, which are heavily formatted.
Allen replied, "I'd hate to think that when I died a computer would write my obituary. Can we at least have somebody spend 15 minutes writing my obituary?"
Want to talk with StatSheet's Robbie Allen? Join us for a live chat with him on Poynter.org at 3 p.m. ET on Wednesday.
Coming Tuesday: How to be in the right place at the right time.