CNBC, By Ash Bennington, Dec 22, 2010, A 2+20 Payout from Trading Tweets?
There's really no middle ground on this one: Trading stocks based on data collected through Twitter is either sheer genius or abject stupidity.
Derwent Capital Markets, a hedge fund, is going to begin trading in February based on just such a model.
Their trading model will be based on such things as "the number of times words on Twitter such as "calm" rise above or below average" According to a paper published jointly by the University of Manchester and Indiana University, "A change in emotions expressed online would be followed between two and six days later by a move in the index, the researchers said, and this information let them predict its movements with 87.6 percent accuracy."
Another press release is Your Tweets Could be Worth Millions.
Here is a sceptical view from Felix Salomon:
This paper, which has now become a fully-fledged hedge fund, is certainly good at hitting buttons: not only does it include Twitter and stocks, but it even finds room to include an important role for Google n-grams!
To my untrained eye, I have to admit that all I see here is two random lines layered on top of each other. But according to the paper, if you run this data through a Granger Causality Analysis and then a Self-Organizing Fuzzy Neural Network, you get all manner of enticing predictive power out the other end. In the chart, the shaded areas supposedly show the periods where Twitter successfully predicted where the stock market was going.