The speed and accuracy of social media sentiment analysis presents barriers to predicting stock price movement, a survey of finance professionals has shown.
The survey conducted by Colt of 360 UK financial sector professionals, including brokers and heads of trading desks, showed that only 7 percent of respondents would consider using social media sentiment as a primary tool for predicting stock price movement.
Many finance professionals, 63 percent, believe that tracking public sentiment on sites such as Twitter can be directly linked to the valuation of individual stocks, meaning that there is some value in the sentiment analysis, even as a trailing indicator.
Companies such as hedge funds are able to scan data from social media sites at random, before categorising these into one of a range of public mood states. Algorithmic trading strategies are then used to place trade orders, potentially giving a competitive advantage over competitors using more traditional trading methods.
However, a third of respondents believe that the speed with which analysis could be made created a stumbling block to adopting social media sentiment analysis.
Part of the problems is the amount of data that needs to be processed, with 43 percent claiming that they would struggle to respond quickly to the influx of information generated by social media sites.
"Data mined from millions of tweets and Facebook posts will only add to the increasingly large volumes of information flowing through a firm's IT systems," said Hugh Cumberland, Solution Manager, Payment & Settlement Services at Colt.
"With additional capacity and bandwidth required to store, access and manipulate the millions of messages, social media analysis will need to be underpinned by an appropriate IT infrastructure to ensure a consistent, fast and reliable flow of data into the heart of the trading environment."
Another concern for finance professionals is the accuracy of data, which respondents believe could prevent the use of sentiment analysis from achieving mainstream adoption.
"Getting data to the heart of the trading systems as fast as possible is nothing new," Cumberland added.
"Addressing anxiety over data integrity requires confidence that the tools can accurately separate credible data from the general social noise along with maliciously generated content."
Dr Daniel Beunza, Lecturer in Management for the London School of Economics (LSE), said that while sentiment analysis presented was an "interesting concept" for traders, its full potential had not yet been reached.
"One approach could be based on trending topics, which is different from sentiment," said Dr Beunza. "Another could be on semantic analysis, which is not just analysing the positive or negative, but the actual content itself."
"However, I'm yet to see any companies adopt this approach yet."
Sign up for Computerworld eNewsletters.