•New

Professor Arratia was commissioned to carry out the research project by the financial consultancy firm Acuity Trading. It consists of an econometric study that determines the capacity of social sentiment indicators developed by the firm to predict financial market movements.

To carry out the study, Professor Argimiro Arratia used a vector auto-regression (VAR) model that measures the relationship of causality between different sets of sentiment indicators and the financial series to be predicted (the history of stock prices, oil prices, dollar exchange rates, etc.). A VAR model is an econometric model that describes the evolution of a given number of variables over a period of time. Sentiment analysis uses natural language processing and computational linguistics to identify and extract information from texts.

The conclusion of the study is that some indicators do have a certain predictive capacity, depending on the financial series analysed. Of the indicators developed by Acuity Trading, the study shows that, in general, those derived from currency exchange offer the best cause-effect relationship. The results also show that the financial volatility indicator is the most reliable means of predicting the behaviour of any financial series sampled daily.

This research project may be essential for financial service providers or brokers who use big data in their business. The project forms part of the UPC’s top-flight research on big data, for which it has attained international recognition.

Acuity Trading is a financial consultancy firm that uses machine learning and massive data processing technology to draw up indicators of social sentiments based on millions of articles from digital news media.