Social Media Mania Gone Wild? Tweets Will Be Used To Make Computer-Guided Investment Decisions

Wednesday, April 27, 2011 20:19
Social Media Mania Gone Wild? Tweets Will Be Used To Make Computer-Guided Investment Decisions

Tags: asset management | Offbeat | Social Media

In one of the stranger manifestations of social media mania, a fund is being created that will invest based on computer-programmed analysis of the social stream. 

This Website Is For Financial Professionals Only


Andrew Graham, a media strategist who specializes in finance, energy, and public policy, says in a post on Mashable that London-based Derwent Capital Markets is planning to launch a  “Twitter hedge fund”  and has raised an estimated $100 million in capital.


Graham suggests the Twitter stream will be analyzed by computers that use some sort of algorithm to determine investment sentiment, but he is short on details about how this would work generate buy and sell signals of securities. An article on Bloomberg last December, however, provides more detail on the strategy.


Bloomberg reported that a paper, entitled Twitter Mood Predicts The Stock Market," by professors from the University of Manchester and Indiana University and published in October 2010 said the number of emotional words on Twitter could be used to predict daily moves in the Dow Jones Industrial Average. Derwent has teamed up with one of the authors of that research paper.


"Here we investigate whether measurements of collective mood states derived from large-scale Twitter feeds
are correlated to the value of the Dow Jones Industrial Average (DJIA) over time," says the paper. "Our results indicate that the
accuracy of DJIA predictions can be significantly improved by the inclusion of specific public mood dimensions but not others.
We find an accuracy of 87.6% in predicting the daily up and down changes in the closing values of the DJIA and a reduction
of the Mean Average Percentage Error by more than 6%."


Could you imagine a fund manager using tweets to make investment decisions? Is this a sign of a top in the use of social media?







Comments (1)

Seems a bit of a stretch, but should be technically possible to pull off. Not to say that it will perform well just because it's possible. While this is the first time I've heard of tweets being used in this capacity, "news based" trading algorithms have been around for a while. You'll see them discussed in hushed tones on electronic forums. Their creators are trying to use keyword analysis to extract the essential information faster than flesh and blood readers.

Ultimately, both day traders and fundamental managers use information contained within text documents as part of their attempts to predict price movement. It tends to be in news stories, press releases, newsletters, and economic releases... but its still just text conveying some sort of meaning that could be analyzed by either humans or machines. if we believe that useful predictions can come from headlines, I don't see why they couldn't come from tweets.

Recently, I've seen several articles and videos where researches from Google, Twitter, and have discussed how various forms of message posting and web searches create clear patterns. For instance, the winners of world cup events could be determined by the patterns of activity in each team's country. Browsers in winning countries would have great search or discussion activity than the losers, and it would be more focused on soccer.

Given that, it sounds extremely difficult to detect and apply appropriate signals, twitter might not hold enough information to drive significant backtesting of their process, and I don't think I'd want to be the compliance officer explaining the method to a regulator if it goes south.
gabriel.cooper , April 28, 2011

Write comment

You must be logged in to post a comment. Please register if you do not have an account yet.