The tone of Twitter posts can help predict the result of football matches when combined with betting market prices, a study has found. Twitter activity can predict when a team is more likely to win and soccer bets are mispriced, according to researchers from University of East Anglia (UEA) in the UK. The study, published in the journal Economic Inquiry, examined 13.8 million tweets – an average of 5.2 tweets per second – during an English Premier League (EPL) season.
These were compared with in-play betting prices available at the same time on online betting exchange Betfair. They found that if the combined tone of tweets in a given second during a match was positive – as measured by a micro-blogging dictionary – then the team was more likely to win than the betting market prices implied. Tweets were particularly informative in the aftermath of goals and red cards, suggesting that social media content is particularly useful in assessing the implications and significance of new information.
Social media is used as a forecasting tool by a variety of firms and agencies, but the researchers wanted to find out how useful and accurate it is. They measured the aggregate tone of all tweets for each team, in each second of 372 matches that took place during the 2013/14 season. “We find that Twitter activity predicts match outcomes, after controlling for betting market prices,” said Alasdair Brown, of UEA’s School of Economics.
“Much of the predictive power of social media presents itself just after significant market events, such as goals and red cards, where the tone of Tweets can help in the interpretation of information. In short, social media activity does not just represent sentiment or misinformation. If sensibly aggregated it can, when combined with a prediction market, help to improve forecast accuracy,” said Brown.
The researchers constructed a number of betting strategies to quantify the degree of mispricing that social media predicts. Using conservative estimates of the commission paid to Betfair, and a strategy of betting when Tweets on a team are positive, a bettor could have earned an average return of 2.28 per cent from 903,821 bets. This compares favourably with average returns of 5.41 per cent across the matches studied.