Generating Comparative Explanations of Financial Time Series

Private and professional investors can easily access large amounts of financial data describing the temporal evolution of the stock prices. Making appropriate decisions about financial activities often entails performing comparative studies to get an increased comprehension of the underlying assets. The aim of this work is to automatically generate summarized explanations of financial stock series based on the most established fundamental indicators. Unlike any previous summary protoform, the newly proposed time series explanations (i) are suited to comparative analyses, i.e., they express a relative strength of the summary claim about a given stock compared to a reference stock cluster, and (ii) are based on a time series embedding representation indicating the level of similarity between different stocks/stock groups in various periods. The preliminary results demonstrate the usefulness and applicability of the proposed approach.