IBM’s Watson is supplying the ESPN fantasy football game with artificial intelligence-powered insights. The advanced Fantasy Insights system originally debuted last year but has been incorporated as a key resource for the 2018 season.
Developers at IBM trained Watson to understand the sport by giving it more than six million articles to read. The platform stays up to date by ingesting 3,000 new sources every hour. While NFL player statistics are already readily available to fantasy gamers, Watson is supplementing those raw numbers with “unstructured data.” Watson culls through more than 10 million news stories, blogs, forums, rankings, and tweets to estimate the potential upside and downside of a matchup, analyze boom or bust chances, predict injury, and convey media sentiment and trends.
For instance, prior to Week 1, Watson projected that Patriots quarterback Tom Brady would throw for 284.3 yards with 2.1 touchdowns and 0.6 turnovers, and thus tally 19.4 points in a standard scoring system. Additionally, the AI platform rated Brady as having a 29 percent “boom” chance (scoring more than 23.8 points) and a 12.2 percent chance of a “bust” (12 or fewer points). In defeating the Texans 27-20, Brady actually threw for 277 yards with three TDs and an interception. That performance equated to 21.3 points.
Similarly, Watson pegged Steelers running back James Conner for 67.5 yards, 0.5 TDs, and 11.1 fantasy points and against the Browns. It also assigned him a 23.4 percent “boom” chance. In Pittsburgh’s 21-all tie with the Browns, Conner ran for 135 yards, two TDs, and 29.2 fantasy points.
Each of Watson’s projections is the result of 1,000 simulations. IBM master inventor and data scientist Aaron Baughman wrote an extensive summary of the training process and how his fantasy team went a perfect 13-0 while using Watson’s Fantasy Insights last year. This season, IBM has organized a league for its employees and some famous sports fans (actor Jerry Ferrara, former NFL player Charles Woodson, and former football player and current mathematician John Urschel). ESPN also is producing weekly advice stories based on Watson recommendations.
“Being able to tap into both structured and unstructured data to allow you to make the most informed decision—it’s applicable to fantasy sports, obviously,” said Noah Syken, VP of sports and entertainment partnerships at IBM, in a statement. “But it’s also applicable to legal and medical and financial trade-offs that people make every single day.”
SportTechie Takeaway
Fantasy sports are a $7.2 billion industry with 59.3 million competitors, according their trade association, the FSTA. Football is the primary driver of those figures, so this is no small undertaking by IBM. Given the rise of daily fantasy sports—which assuredly soon will be joined by a growth in sports betting—the market for additional sports intelligence and insights definitely has a much higher chance of boom than bust.