Economists have long pondered the correlation between stock market performance and the mood of investors, but have had difficulty proving economic significance and transforming it into a profitable strategy.
However, big data may be the answer to how we can use sentiment measurements to improve investment decisions.
On Thursday, November 24, 2016, 200 professionals from the Luxembourg financial community attended lux future lab’s third installment of its FinTechMeets series. The event, entitled Big Data & Emotions in Asset Management, considered the possibility and challenges of capturing and interpreting emotions, specifically through social media channels.
After an introduction by Karin Schingten, CEO, lux future lab, Yves Nosbusch, Chief Economist of BGL BNP Paribas, presented on the field’s most striking findings, behavioral models and recent empirical research.
According to classical finance or, more broadly, economics, emotions hold no weight. Despite evidence of an emotional component to investment decisions, classical investors disregard it as irrelevant as long as there are enough rational investors to keep prices in line with fundamental values.
Past studies have proven that football losses and weather, for example, impact stock prices, which is at odds with the traditional view that prices are driven exclusively by factors fundamental to the value of the company.
Nosbusch also touched upon the price volatility noise traders can create, as well as the riskiness of real-world arbitrage, concluding that there are solid reasons to believe that sentiment and emotions significantly affect asset prices.
About 10 years ago, more sophisticated methodologies made it possible to extract information from large data sets, such as media content. Findings showed the tendency for those basing trading decisions on stale news to irrationally react to what had already been impounded in the market price.
While economists have made statistically significant findings, large enough data sets, improved processes and economic significance are required to elicit value from those findings through a profitable trading strategy.
SESAMm Cofounder and President Sylvain Forté explained how his company is working to do just that by extracting big data, quantifying emotions, building predictive models and deciphering valuable information for investment professionals.
Big data, which is defined by its high levels of volume, variety and velocity, and big-data tools give access to new market information that plays a role in driving market prices.
Asset management, he reminded, is fundamentally a big-data task, with billions of data points moving in real time from various sources. With modern tools, the ability to generate alpha based on textual data is in reach.
Forté noted that primitive global emotions — fear, sadness, anger, joy, disgust and surprise — are highly related to trading sentiment and can also be used to measure a company’s global emotional profile, i.e. the feelings its brand inspires.
Due to the complexity of word meaning, extracting emotions from text is no simple matter, requiring the proper understanding of irony, word sense, negation and point of view.
The future of big data and the utilization of emotions to predict asset prices are closely intertwined and, despite the challenges, their shared outcome looks promising and highly advantageous.