Algorithms that depend on machine readable news techniques have caused a minor splash in the trading world. A lot of these products scour the Internet for news articles and extract keywords that are fed into a model that can help determine public sentiment.

Two professors of informatics and computing at Indiana University, Johan Bollen and Huina Mao, have narrowed their focus to Twitter posts and have found that "collective mood states derived from Twitter feeds" like OpinionFinder and Google-Profile of Mood States can predict the Dow Jones industrial average.

Hedge fund manager Paul Hawtin of Derwent Capital Markets liked the idea so much that he's purchased exclusive rights to the model. He plans to dedicate an entire hedge fund to it and has apparently hired the two professors as consultants to build out a formal trading model. Bollen and Mao reportedly will take home 35 percent of any income received from the University's licensing contract with Derwent Capital.

This looms as an interesting test case. Most likely, the actual implementation will be harder than the research. To some, this comes across as the height of folly, surely a sign that the Twitter revolution has peaked and perhaps a sign that algorithmic trading has gotten a bit too elastic. But the proof will be in the numbers. You can bet people will be carefully monitoring returns. My sense is that this will be a hard sell to potential limited partners.  

For more:
- here's the paper, "Twitter mood predicts the stock market" (.pdf)
- here's a Huffington Post article

Related Articles:
Hedge funds to profit from Twitter feeds
Citi embraces Twitter as customer service tool
Banks and Twitter: Compliance issues