Investors have turned to astrologers, palm readers and even
economists to help them predict the stock market. Now they can add to
this list the latest darling of the Internet, Twitter.
Yes, that message board where you have to communicate in 140
characters or less is being touted by an Indiana University
professor as the next big thing when it comes to predicting stock
market performance with 90% accuracy.
Johan Bollen, an associate professor at the Indiana University
School of Informatics and Computing, and two colleagues claim that
Twitter is an accurate predictor of mood, and that the level of
relative calm or anxiety in the Twitter-verse roughly matches the
performance of the Dow Jones Industrial Average, four days ahead of
time.
Bollen conducted his research with his Indiana University
colleague Huina Mao and Xiao-Jun Zeng at the school of computer
science at the University of Manchester. They analyzed millions of
tweets in 2008 and sorted them according to mood.
They looked at whether the tweeters were happy or sad, calm or
anxious, hostile or agreeable, clearheaded or confused? The
researchers then plotted the fluctuating moods against a graph of
the 2008 Dow Jones Industrial Average. They say some mood pairs
such as happy/sad, or clearheaded/confused didn't have predictive
power, but the calm/anxious pairing did.
Their report isn't the first one to identify a link between stock
market performance and public mood, but using social media to do so
is a new wrinkle.
Some critics will debunk the studies as data mining and picking
information that creates a pattern and analyzing the results after
the fact. They dispute whether it would work in real time and even if
it does, it might still be difficult to make much money with this
information.
Bollen of course defends the research and calls the results
"pretty convincing" and potentially lucrative. He believes
a lot of money can be made and adds that a number of hedge funds are
already interested in his findings.
Investors have turned to astrologers, palm readers and even economists to help them predict the stock market. Now they can add to this list the latest darling of the Internet, Twitter.
Yes, that message board where you have to communicate in 140 characters or less is being touted by an Indiana University professor as the next big thing when it comes to predicting stock market performance with 90% accuracy.
Johan Bollen, an associate professor at the Indiana University School of Informatics and Computing, and two colleagues claim that Twitter is an accurate predictor of mood, and that the level of relative calm or anxiety in the Twitter-verse roughly matches the performance of the Dow Jones Industrial Average, four days ahead of time.
Bollen conducted his research with his Indiana University colleague Huina Mao and Xiao-Jun Zeng at the school of computer science at the University of Manchester. They analyzed millions of tweets in 2008 and sorted them according to mood.
They looked at whether the tweeters were happy or sad, calm or anxious, hostile or agreeable, clearheaded or confused? The researchers then plotted the fluctuating moods against a graph of the 2008 Dow Jones Industrial Average. They say some mood pairs such as happy/sad, or clearheaded/confused didn't have predictive power, but the calm/anxious pairing did.
Their report isn't the first one to identify a link between stock market performance and public mood, but using social media to do so is a new wrinkle.
Some critics will debunk the studies as data mining and picking information that creates a pattern and analyzing the results after the fact. They dispute whether it would work in real time and even if it does, it might still be difficult to make much money with this information.
Bollen of course defends the research and calls the results "pretty convincing" and potentially lucrative. He believes a lot of money can be made and adds that a number of hedge funds are already interested in his findings.
