What You Say and How You Say It Matters: Predicting Stock Volatility Using Verbal and Vocal Cues

Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics, pages 390–401 Florence, Italy, July 28 - August 2, 2019. c⃝2019 Association for Computational Linguistics 390 What You Say and How You Say It Matters: Predicting Financial Risk Using Verbal and Vocal Cues Yu Qin School of Information Renmin University of China qinyu.gemini@gmail.com Yi Yang ∗ HKUST Business School Hong Kong University of Science and Technology imyiyang@ust.hk Abstract Predicting financial risk is an essential task in financial market. Prior research has shown that textual information in a firm’s financial state- ment can be used to predict its stock’s risk level. Nowadays, firm CEOs communicate information not only verbally through press releases and financial reports, but also non- verbally through investor meetings and earn- ings conference calls. There are anecdotal evidences that CEO’s vocal features, such as emotions and voice tones, can reveal the firm’s performance. However, how vocal features can be used to predict risk levels, and to what extent, is still unknown. To fill the gap, we ob- tain earnings call audio recordings and textual transcripts for S&P 500 companies in recent years. We propose a multimodal deep regres- sion model (MDRM) that jointly model CEO’s verbal (from text) and vocal (from audio) in- formation in a conference call. Empirical re- sults show that our model that jointly consid- ers verbal and vocal features achieves signif- icant and substantial prediction error reduc- tion. We also discuss several interesting find- ings and the implications to financial markets. The processed earnings conference calls data (text and audio) are released for readers who are interested in reproducing the results or de- signing trading strategy. 1 Introduction Predicting financial risks of publicly traded com- panies is of great interest to capital market partic- ipants. In finance, stock price volatility, which is the standard deviation of a stock’s returns over a period of time, is often used as a measure of finan- cial risks. Unlike directly predicting stock prices, it is uncontroversial in the field of economics that one can predict a stock’s volatility level using pub- licly available information (Bernard et al., 2007). Based on this assumption, a burgeoning body of ∗Corresponding author. research, both in finance and computational lin- guistics, has studied predicting stock volatility us- ing various textual sources, including company disclosed reports (Kogan et al., 2009), public news articles (Tetlock, 2007), company earnings call transcripts (Wang and Hua, 2014), and social me- dia (Ding et al., 2015). Thanks to technological advances, massive amounts of unstructured multimedia data, such as investor conference audio records and CEO pub- lic speech videos, have been archived and can be accessed by institutional and individual investors. Everything CEOs (or other executives) say will be closely examined and analyzed by investors. There are anecdotal evidences that CEO’s nonver- bal features, such as emotions and voice tones, can also be used to reveal firm’s performance. For ex- ample, it has been reported that hedge fund com- panies hire ex-CIA agents trained in reading non- verbal …

Content truncated to 500 words