Can tweets predict NBA hoop dreams for game-day strategy? Yes, says RIT expert

Saunders College professor conducted research study on sentiment and sports performance

Follow Marcia Morphy on Twitter
Follow RITNEWS on Twitter

Photo provided

Yang Yu

What if tweets by NBA players the day of the game could predict their performance on the court?

Yang Yu found a direct correlation during his data-driven analytics on the Twitter social media website. After tracking tweets by more than 200 NBA players during the past two years, he said there is a positive correlation between sentiment and sports performance.

“If players’ tweets on the day of the game are positive they are more likely to play better than if their comments on Twitter are negative,” said Yu, assistant professor of management information systems in Saunders College of Business at Rochester Institute of Technology.

“My research shows it works,” said Yu about “Hidden In-Game Intelligence in NBA Players’ Tweets,” which is published in the November issue of Communications of the ACM. Chun-Keung (Stan) Hoi, RIT finance professor in the Saunders College, and Chenyan Xu, computer science and information systems professor at Richard Stockton College of New Jersey, were research collaborators for the study.

Yu said the past few years have seen an explosion of tweets by NBA players who have used this emerging technology, direct messages of 140 characters or less, to communicate with fans, journalists, peers, friends and others.

Key insights from the study are revealing:

  • Sentiment analysis, an emerging text-mining technique, can help analyze the tweets of NBA players.
  • Coaches, general managers, and other staff of NBA teams can use it to discern and address players’ moods before games.
  • The study shows players’ before-game mood, as captured through the sentiment they reveal in their tweets, is positively associated with their on-court performance.

The correlation is also influenced by other factors. For example, in away games, players’ performances are more likely to be affected by their sentiments, and this effect is stronger for players who earn higher salaries.

Yu cited Kevin Durant, one of the NBA’s Oklahoma City Thunder and the league’s Most Valuable Player (2013-2014) who has more than 8.2 million followers on Twitter and has sent out more than 22,000 tweets.

In a tweet by Durant on Feb. 14, 2013, he said, “I’m aight (sic) after, thanks for asking…mad about the loss, but headed to Houston to enjoy my 4th all star game!!”

As Yu explained, “Durant is a good example of the connection between social media expression and performance. This type of social media analytics research can be beneficial to coaches and managers in determining game-day strategy.”

Yu also says his research and text-mining approach can be applied to professional athletes in other sports. He recently completed a collaborative study on the “World Cup 2014 in the Twitter World,” which was published in Elsevier Journal, Feb. 2015.