Can Twitter Predict the Future?

Today a hedge fund announced that it was going to start tracking stocks via Twitter.  The hedge fund, The Derwent Absolute Return Fund Ltd. in London, will use an algorithm that will follow updates on the social network and use trending to determine what and when to trade.

It’s an interesting experiment, and I assume the fund has been practicing their algorithm against fake money or their own money for a while now, but it will be interesting to view the returns from the fund.  My thought on trying this is that if we start looking at who is contributing to the twitter ecosystem, can Twitter really predict the stock market, or for that matter any other event?

Yesterday Sysomos released a study done on Twitter demographics and the way users are using the service.  Here are some of the highlights:

  • Users with 100+ friends have increased by three-fold to 21% since 2009.
  • 22.5% of users accounted for about 90% of all activity.
  • 80% users have made fewer than 500 tweets.
  • Justin Bieber is one of top two-word phrases and top name in user’s bios.
  • Significantly more users are disclosing their location, bio and web information in Twitter profiles.

If Justin Bieber is giving out free stock advise, then the hedge fund is sure to score!  I am impressed that users are following more people, but unless these users are creating content (original tweets or retweets), it may not help develop the algorithm.  Of course, you can have the opposite problem where spammers contribute too much content that screws with the trending.

In the experience we had with trending on AOL Lifestream we had to constantly manipulate the algorithm as spammers and general users made trends at times irrelevant.  In a later post, I am going to delve more into trending and the challenges it presents and whose social data may be most relevant in contributing to interesting trends.

Leave a Reply

Your email address will not be published.