Will Duff Gordon is editor of DataExplorers, an analyst firm which digs in to stock market trends.
Stock markets are skittish and here is why: in 10 weeks, 2011 has so far thrown up three revolutions. At a time of tentative emergence from recession this inevitably leads to “jumpy” investor behavior.
Is there anything one can “model” from this? “Doing nothing” is hardly an impressive way to react from active fund managers whose fees were laid bare by low fee czar, John Bogle (Vanguard founder) as $69 billion now compared to $12 billion in 1990.
The quantitative fund managers have had a tough time but hope returns if they can systematically make use of crowd feelings since it has been shown that Twitter feeds can predict the Dow Jones (NSDQ: NWS) with over 80 percent accuracy on a three to four day lag. Let’s look behind the numbers.
“Twitter mood predicts the stock market” (Johan Bollen, Huina Mao, Xiao-Jun Zeng) is a paper by academics from the Universities of Indiana and Manchester. They used 10 million Tweets from 2.7 million people during 2008 as their sample and two mood statement analyzing tools.
The bottom line is that the combination of happy and calm crowd sentiment is rather good at predicting the direction of the Dow Jones stock market. Why is this interesting? If the daily chopping and changing of the stock markets is linked to the emotional state of the masses it is as well to use this factor for funds with daily trading strategies, at least as an overlay.
The concept of reading crowd sentiment is not new. What is new is the idea to use Twitter rather than more complicated engines offering machine readable news.
What is also new is the understanding of the important role played by social media, including Twitter, in spreading public opinion in the run up to Tunisia and Egypt’s revolutions. The historian in me would like to challenge this, since it could be that these firms are merely associating themselves with popular uprising without due basis. But at the very least, it is clear that Twitter et al are major organs of public opinion.
The other impressive aspect is that this predictive power takes place both when Twitter is quite a small outlet for the mass mood (2.7 million Tweeters is not a big sample) along with the turbulent period over which they ran the study.
On this point, note the sobering fact that, “the deviation between Calm values and the DJIA on that day (when the Fed agreed a massive bailout plan on October 13th) illustrates that unexpected news is not anticipated by the public mood yet remains a significant factor in modeling the stock market.”
Since a tweet cannot be bigger than 140 characters, machines (as used in this study) such as Opinion Finder and Google-Profile of Mood States have a better chance of categorizing the mood expressed. Products like the Reuters (NYSE: TRI) NewsScope Analytics attempt to turn breaking news articles into sentiment but face the obvious challenge of placing ambiguous news (like company results that beat estimates but give a negative outlook) into tight buckets. Twitter doesn’t present this problem.
Taking the temperature of crowds is worth the attention this is getting, given the daily struggle that quant funds face in managing volatility. On the other hand, politicians could be even more interested!
Next time you feel anxious or happy, tweet – someone somewhere might find it rather useful……
This article originally appeared in DataExplorers.