A Collection of Articles
Edit
Emerging Technology from the arXiv

A View from Emerging Technology from the arXiv

Twitter Mood Predicts The Stock Market

An analysis of almost 10 million tweets from 2008 shows how they can be used to predict stock market movements up to 6 days in advance

  • October 18, 2010

There’s no shortage of people who say they know how to predict whether the stock market will go up or down on a particular day. But there are few, if any, who can do it consistently better than tossing a coin.

For many economists that’s easy to explain. Conventional economic theory holds that the movement of prices in a perfect market should follow a random walk and should be impossible to predict with an accuracy greater than 50 per cent.

There’s a fly in this economic ointment, however. Numerous studies show that stock market prices are not random and this implies that they ought to be predictable. The question is how to do it consistently.

Today, Johan Bollen at Indiana University and a couple of pals say they’ve found just such a predictor buried in the seemingly mindless stream of words that emanates from the Twitterverse.

For some time now, researchers have attempted to extract useful information from this firehose. One idea is that the stream of thought is representative of the mental state of humankind at any instant. Various groups have devised algorithms to analyse this datastream hoping to use it to take the temperature of various human states.

One algorithm, called the Google-Profile of Mood States (GPOMS), records the level of six states: happiness, kindness, alertness, sureness, vitality and calmness.

The question that Bollen and co ask is whether any of these states correlates with stock market prices. After all, they say, it is not entirely beyond credence that the rise and fall of stock market prices is influenced by the public mood.

So these guys took 9.7 million tweets posted by 2.7 million tweeters between March and December 2008 and looked for correlations between the GPOMS indices and whether Dow Jones Industrial Average rose of fell each day.

Their extraordinary conclusion is that there really is a correlation between the Dow Jones Industrial Average and one of the GPOMS indices–calmness.

In fact, the calmness index appears to be a good predictor of whether the Dow Jones Industrial Average goes up or down between 2 and 6 days later. “We find an accuracy of 87.6% in predicting the daily up and down changes in the closing values of the Dow Jones Industrial Average,” say Bollen and co

That’s an incredible result–that a Twitter mood can predict the stock market–but the figures appear to point that way.

Is it really possible that the calmness index is correlated with the stock market? Maybe. Back in April we looked at some work showing how tweets about films can be used to predict box office takings.

But there are at least two good reasons to suspect that this result may not be all it seems. The first is the lack of plausible mechanism: how could the Twitter mood measured by the calmness index actually affect the Dow Jones Industrial Average up to six days later? Nobody knows.

The second is that the Twitter feeds Bollen and co used were not just from the US but from around the globe. Although it’s probably a fair assumption that a good proportion of these tweeters were based in the US in 2008, there’s no way of knowing what proportion. By this reckoning, tweeters in Timbuktu somehow help predict the Dow Jones Industrial Average.

Either way, this work is bound to attract interest. And taken at face value, it could be hugely influential. If calmness does have real predictive value of the stock market, we’ll see an explosion of interest in financial Twitter analytics. And Bollen and co should soon become extremely wealthy individuals.

Ref: arxiv.org/abs/1010.3003: Twitter Mood Predicts The Stock Market

Uh oh–you've read all five of your free articles for this month.

Insider basic

$29.95/yr US PRICE

Subscribe
What's Included
  • 1 year (6 issues) of MIT Technology Review magazine in print OR digital format
  • Access to the entire online story archive: 1997-present
  • Special discounts to select partners
  • Discounts to our events

You've read of free articles this month.