A panel with Howard Lindzon of StockTwits, Johan Bollen at Guidewave Consulting, Fahad Kamr at Market IQ and Richard Tibbetts of StreamBase having a discussion on why hedge funds, banks and financial services need social data.
A key component of this panel centered on how traders were consuming social data – machine learning vs human consumption. Howard Lindzon kicked off the conversation by comparing social data to the ‘ticker tape’ of stock trading that existed in the 1920s. People would read the orders and watch the tape as a means of gaining insider information. Social data, as Howard sees it, is the new, instant (and legal!) version of the tape.
Tellingly, Howard takes a personal perspective on trading, as opposed to the machine approach, although the opportunities for algorithmic consumption seem clear. As he said, “We have to learn who we are. We’re not Wall Street. StockTwits job is to give users a way to express themselves financially.” And to his point, the human nature of social data consumption is one that is growing in the market. Seth McGuire of Gnip talked about how traders at investment banks hide their iPhones under their desks because they want to digest the insight coming from Twitter streams, but aren’t allowed by compliance officers to access that data on their desktop.
In defense of machine learning, Johan Bollen discussed his research at Indiana University School of Informatics and Computing. His team focuses on sentiment analysis in social data streams and how overall changes in sentiment can predict the market. His enthusiastic defense of the wisdom of the crowd delved into how algorithms can learn and adapt to changes in that crowd data. In responding to that point, Richard Tibbetts of Streambase raised some interesting questions on the lifespan of alpha within a stream of this nature. It is the nature of an efficient market to correct for alpha – why would a Twitter stream be different?
Fahad Kamr of Market IQ talked about how the newness of social data in the financial world means that “No one really knows what’s up with it”. In the context of Richard’s point on alpha, it will be interesting to see what uses cases retain alpha (sentiment? news-driven?). News-driven and event-trading was particularly discussed, especially in relation to current news and sentiment streams from traditional finance data providers. User reliability will be an important consideration in comparision, as Fahad noted. How do you separate the noise (including those who might try to game the system)?
What the industry requires is actionable insight and a pattern of that insight across time – a common theme across Big Boulder, with interesting discussions on how attendees, panelists and platforms themselves were attempting to provide said insight.