Trade the Event: Corporate Events Detection for News-Based Event-Driven Trading

Findings of the Association for Computational Linguistics: ACL-IJCNLP 2021, pages 2114–2124 August 1–6, 2021. ©2021 Association for Computational Linguistics 2114 Trade the Event: Corporate Events Detection for News-Based Event-Driven Trading Zhihan Zhou Liqian Ma Han Liu Department of Computer Science, Northwestern University {zhihanzhou, liqianma}@u.northwestern.edu hanliu@northwestern.edu Abstract In this paper, we introduce an event-driven trading strategy that predicts stock movements by detecting corporate events from news arti- cles. Unlike existing models that utilize tex- tual features (e.g., bag-of-words) and senti- ments to directly make stock predictions, we consider corporate events as the driving force behind stock movements and aim to profit from the temporary stock mispricing that may occur when corporate events take place. The core of the proposed strategy is a bi-level event detection model. The low-level event detec- tor identifies events’ existences from each to- ken, while the high-level event detector in- corporates the entire article’s representation and the low-level detected results to discover events at the article-level. We also develop an elaborately-annotated dataset EDT for corpo- rate event detection and news-based stock pre- diction benchmark. EDT includes 9721 news articles with token-level event labels as well as 303893 news articles with minute-level times- tamps and comprehensive stock price labels. Experiments on EDT indicate that the pro- posed strategy outperforms all the baselines in winning rate, excess returns over the market, and the average return on each transaction. 1 1 Introduction By shaping investors’ perceptions and assessments of companies, financial news has significant im- pacts on the stock market (Engle and Ng, 1993; Tetlock, 2007). News-based stock prediction mod- els are thus developed to automatically discover signals of stock market movements from the count- less news articles that generated every moment. (Kalyani et al., 2016; Shah et al., 2018; Mohan et al., 2019). 1Code and the EDT dataset is available at https:// github.com/Zhihan1996/TradeTheEvent Time Ticker: NP Event: Stock Repurchase Event: Dividend Increase Trading ALPHARETTA, Ga., Nov. 26, 2019 – Neenah, Inc. (NYSE: NP) announced today that its Board of Directors approved a four percent increase in the cash dividend on the company’s common stock. The company also announced authorization of a repurchase program. Dividend Increase Stock Repurchase (Trading Policy) (Event Detector) Investing News Articles Figure 1: Overview of the event-driven trading strategy Previous studies mainly rely on textual features and sentiment analysis to forecast the stock move- ments (Fung et al., 2003; Liu, 2018; Huynh et al., 2017). Both of them, however, often face the prob- lem of poor explainability and low signal-to-noise ratio. Textual feature-based methods often formu- late the stock prediction as a text classification problem by directly predicting the rise and fall of stocks based on the extracted features. These mod- els fail to make reasonable trading decisions since they omit the reasons behind stock price changes. Sentiment-based methods avoid this problem by regarding the news articles’ sentiments as the indi- cator of stock movement. However, news sentiment is subjective, which can be greatly affected by the author’s standpoint and writing style. Unlike textual features and …

Content truncated to 500 words