THE INSTANT MESSAGE.

Heather McKenzie examines whether the data gleaned from social media is changing the face of trading.

The frenzy of interest that surrounded the initial public offerings (IPOs) of Facebook and Twitter brought social media into the mainstream and to the attention of financial firms. What started out as an online service aimed at the young has rapidly stretched its tentacles into many other areas.

In its IPO prospectus, Facebook claimed 845 million active monthly users and a website that featured 2.7 bn daily ‘likes’ and comments. On May 18 2012, Facebook’s IPO valued the company at over $104 bn. More than a year later, on 7 November 2013, Twitter was sitting on a market cap of around $31bn. That autumn saw the company grow to 200 million users who sent more than 400 million tweets (messages of 140 characters or less) per day. Nearly 60% of tweets are sent from mobile devices.

In 2012 hedge fund manager Derwent Capital attempted to tap into the social media world by setting up a ‘Twitter hedge fund’, the Derwent Capital Markets Absolute Return fund. However it operated for only one month although it generated a 1.86% return in its sterling shares, which according to the Financial Times, was ahead of the market and the average hedge fund.

The fund examined tweets from users of Twitter, as well as messages posted on Facebook and other social media, and created measures of sentiment towards individual stocks, wider markets and commodities. Derwent claimed moves in sentiment gave a lead of about three days on actual price moves. It is now launching a system for private investors that overlays its Twitter-derived measures of sentiment on the IG Group’s spread-betting trading platform.

It is likely to be one of many as a survey conducted by UK-based telecoms company Colt into social media and trading during October 2012 found that 63% of the 360 finance professionals polled believed the performance of individual stocks could be directly linked to public sentiment contained in social media channels. However, lack of confidence in the accuracy of information was the biggest barrier to employing trading algorithms based on sentiments and topics exchanged via these channels. Further, 43% of respondents felt traders would struggle to handle the increasing volumes of data that would be generated by social media sentiment analysis.

Hashtag trading

According to industry analysts Tabb Group, emerging techniques and technologies are providing a more serious use case for social media, enabling it to be used as a trading tool by every level of trader, from investment professionals using algorithms to day traders and retail investors. In a report, Social Alpha: Channelling the Chatter, authors Adam Sussman and Valerie Bogard say start-up firms as well as traditional market data providers such as Bloomberg and Thomson Reuters are offering better social media analytics with the ability to automate incorporation of social media data into trading amidst the continuing growth in quantity and types of social data.

Thomson Reuters’ offering, launched in March 2012, is an extension of its machine-readable news service, Thomson Reuters News Analytics. It searches social media and blog content to deliver analytics on selected companies and market segments and is designed to help trading and investment firms to identify and capitalise on new opportunities. At launch it provided access on up to 50,000 news sites and 4 million social media sites.

In August 2011, Bloomberg began distributing sentiment data for consumer stocks and airlines via its data feed and an arrangement with social market analytics firm WiseWindow. WiseWindow’s service predicts market trends based on customer sentiment expressed on social media sites.

While investors have been using social media sites for years as part of their decision-making process, the huge volumes of data from services such as Twitter prompted Tabb Group to investigate whether such data can be analysed effectively and accurately. “Many advances have been made in this field and we are beginning to see some of the companies developing analytics tools that can find signals for trading, but it requires a significant effort to do this,” says Bogard. “The key to success lies in having platforms that can limit the data to what you want to receive.”

She says several of the tools being developed have their origins in news analytics platforms. There are, however, difficulties in drawing out sentiment from social media, particularly in terms of its context. “It is often difficult to attach meaning to data, particularly the very short posts that can be distributed on Twitter, for example.” Further, sarcasm and the use of emoticons also prevent clarity for those attempting to glean meaning from social media.

Fact from fiction

In analysing social media data, the first challenge is determining what content should be included, says Tabb Group. Techniques include the links inside tweets, which could lead to any number of secondary pieces of content, says Bogard, pointing out that once the data has been scrubbed and filtered, the most common type of analysis is sentiment. “The most basic form is polarity, a negative or positive sentiment. Richer sentiment analysis can include a wide range of emotions, such as frustration, fear or joy. An even more elaborate analysis can go beyond emotions and describe broader macroeconomic themes and concepts. And then it gets even more complex.”

Tabb Group argues that the sheer number of social analytic start-up firms focusing on the space shows the level of interest for this emerging technology. However, the trend is still in its infancy and there are a variety of obstacles that both vendors and users face. Bogard cites research undertaken by one trading firm that found that although many respondents expressed an interest in social media more than half were not yet using it in their trading strategies. Moreover, many were very sceptical about social media, particularly Twitter. “Many investment professionals regard social media as an entertainment channel and repository of videos of cats and kittens,” says Bogard. “They don’t regard it as a vital source of information about the capital markets.”

Hugh Cumberland, business development manager at Colt, says data historically used in the trading world has been kept in relational databases, is highly structured, formatted and cross-referenced, which gives it great integrity. Analysing such data via query languages such as SQL also delivers clear, understandable results. “The explosion of social media data has led to some firms wanting to mine that information for their trading activities,” he says. “But this data is unstructured, is not validated and is in a free format.”

Another significant challenge for social media data is its lack of credibility, he adds. Groups have recently hijacked Twitter accounts, causing significant share price movements via ‘pump and dump’ scams.

Still a burgeoning field, improvements need to be addressed, says Bogard. Despite significant progress in analysing English language social media, analysis of other languages, including internet slang associated with social sites, is under-developed, severely limiting applicability. But according to Bogard’s co-author Sussman, social media is already changing the way news, data and opinions are generated, disseminated and consumed. “For better or worse, financial markets need to adjust accordingly,” he says.

Momentum is building behind the use of social media in the business world and it is reasonable to conclude, despite early failures in social media funds, that it will find its way into the trading world as well. In April 2013, the Securities and Exchange Commission (SEC) cleared the chief executive of video streaming company Netflix of any wrongdoing in using social media platforms to disseminate corporate information. In its ruling the SEC said postings on sites such as Facebook and Twitter are as legitimate as news releases and company websites, as long as the companies inform investors they intend to use such outlets.

“An increasing number of public companies are using social media to communicate with their shareholders and the investing public,” the SEC said in its report. “We appreciate the value and prevalence of social media channels in contemporary market communications, and the commission supports companies seeking new ways to communicate.”

©Best Execution 2014

[divider_to_top]