Most televised sports broadcasts feature advertisements for Fan Duel and Draft Kings, online sites offering daily and weekly fantasy games in football, baseball and other sports. Fantasy sports sites are exempt from prohibitions on online gambling. Federal law treats them as games of skill, not games of chance.
Sports columnists who have written about online fantasy sports say the skills needed to win are not those of everyday fans, but advanced analytics and algorithms — skills taught to MBA students, and to Ph.D. students in economics and mathematics.
Analytics are becoming increasingly important in professional sports. MIT now sponsors an annual conference on sports analytics. Earlier this year, Pittsburgh Tribune-Review sportswriter Travis Sawchick published “Big Money Baseball,” a book in the tradition of Michael Lewis’ “Moneyball.” Sawchick shows how the Pittsburgh Pirates, a low-budget franchise, used advanced analytics to turn around a team that had a 20-year losing record. Pirates manager Clint Hurdle and his staff have a database with millions of data points, on every pitch and swing of the Pirates and their opponents. The Pirates use this data to build lineups, for in-game strategy, and to make decisions on the draft and free agency. This year, the Pirates are one of the top three teams in baseball.
These analytical techniques have also made their way into political campaigns. Consulting firms staffed by MBAs and Ph.D.s offer candidates and political parties “microtargeting” strategies, which seek to identify likely supporters at the individual, rather than the zip code or neighborhood, levels.
One of the most important tasks for any campaign is voter mobilization and turnout. Candidates want to identify likely voters, convert persuadable voters, and make sure all these voters actually cast a ballot. In primary elections, when fewer than one-quarter of registered voters typically go to the polls, identifying supporters is a key to victory.
Traditionally, mobilization relied on a few sources of data — public records showing who had voted in previous elections, and demographic data collected at the zip code level. The demographic data was correlated with polls to determine which groups were most likely to support a particular candidate. Campaigns would then use mailings and phone calls to communicate to all voters in a geographical area with certain demographic attributes.
Microtargeting takes advantage of two developments in the tech world over the past decade. First, for many voters, especially younger people, traditional broadcast and print media have been replaced as sources of political information by online news sites, alternative media, and social media. Second, the reduction in the cost of digital storage has allowed the collection of a very large number of data points on every individual, relating to consumer behavior, political preferences, media habits and many other attributes.
Microtargeting firms use algorithms they have developed over the years to discern patterns of relationships between individual-level consumer, media and demographic data and voting behavior. They then use these algorithms to target political communications, often delivered online, to extremely small subsets of voters.
For example, microtargeting could identify a female voter who lives just west of Boston, owns a condo, drives a hybrid, has a post-graduate degree, works in the health care field, is not married, reads the Boston Globe online, sings in a church choir, and plays in a women’s soccer league, and target a message to her accordingly.
Microtargeting has not been used much in Vermont in previous election cycles, in part because it is very expensive. However, 2016 gubernatorial candidates such as Matt Dunne and Bruce Lisman, who will have well-funded campaigns, may find it worthwhile to use these techniques in seeking to identify likely supporters in what could be a low turnout primary election in August 2016.
Eric L. Davis is professor emeritus of political science at Middlebury College.