Most people know my opinion about twitter – it is the intellectual equivalent of yelling things from a moving car. As such this attempt to to use twitter to track market sentiment doesnt thrill me too much. As an old stats tutor used to constantly drum into our not very bright heads – correlation does not equal causation
THE trend is your friend, as they say on Wall Street. But when it comes to financial decisions, can you trust a Twitter trend? Possibly: an analysis of sentiments expressed on Twitter appears to have given the small London-based firm Derwent Capital Markets an edge.
Derwent’s 25 million fund finished its first month of trading in July with a return of 1.85 per cent. By contrast, the Standard & Poor’s 500 financial index fell 2.2 per cent and the average hedge fund made only 0.76 per cent.
Beginner’s luck? Perhaps. Or maybe the tweets are helping. Derwent’s system tracks emotions expressed across 10 per cent of the roughly 100 million daily tweets using algorithms devised by Johan Bollen, a computer scientist at Indiana University Bloomington. It then uses this information to predict changes in the stock market.
In a study published last year, Bollen’s algorithms predicted the direction of the daily swing of the Dow Jones closing price with 87.6 per cent accuracy. The index consistently rose a few days after a period of “calm” tweets and dipped a few days after a period of “anxious” tweets (arxiv.org/abs/1010.3003).
From New Scientist
