CMStatistics 2021: Start Registration
View Submission - CFE
Title: Forecasting option returns with news Authors:  Gang Li - The Chinese University of Hong Kong (Hong Kong) [presenting]
Bing Han - University of Toronto (Canada)
Jie Cao - Chinese University of Hong Kong (Hong Kong)
Ruijing Yang - The Chinese University of Hong Kong (Hong Kong)
Xintong Zhan - The Chinese University of Hong Kong (Hong Kong)
Abstract: The aim is to study whether text data contains useful information to forecast the cross-sectional equity option returns. We apply both lexicon-based and machine-learning approaches to extract quantitative signals from over six million news articles. The machine-learning methods outperform lexicon-based approaches in predicting the delta-hedged option returns, and generate sizable profits. The predictability is robust after controlling for volatility-related information and other known predictors. An analysis of the keywords identified by machine-learning methods suggests the predictability is largely related to sentiment. We highlight the importance of analyzing unstructured data like texts with machine learning approaches by examining the derivatives market.