Thursday, September 20, 2012

Social Media Predicting Stock Prices? Say What!


No one can dispute that social media is big, but is it big enough to predict stock market prices? Many traders are saying yes and the rise of start-ups offering financial sentiment analysis services based on social media platforms like Twitter are more than confident that it can.

Companies like SNTMNT,  HedgeChatter, DCM Capital and CrowdIndex (still in BETA mode) claim to be able to classify investor sentiment but how accurately can one classify emotions like fear and greed using algorithms? Pretty accurately actually as Johan Bollen, a computer scientist at Indiana University Bloomington, discovered. In a study published in 2010 Johan investigated whether measurements of specific moods derived from large-scale Twitter feeds correlated to the value of the Dow Jones Industrial Average (DJIA) over time. The algorithms he devised actually predicted the direction of the Dow Jones closing price within 87.6% accuracy. His sentiment analysis measured mood in six dimensions (calm, alert, sure, vital, kind and happy). He discovered that the index rose over a few days of "calm" tweets and dipped after a few days of more "anxious" feeling tweets.



Pretty neat huh? The question is how long will it take for social media platforms to not only predict stock prices but actually begin to influence them? 

2 comments: