Do Words Matter: Predicting IPO Performance from Prospectus Sentiment

Abstract

Initial Public Offerings (IPOs) are an important aspect of an investor's portfolio. Due to the incredible amount of risk and uncertainty surrounding them, IPOs make a good investment for investors who are looking for high-risk, high-reward stocks. Investors have typically used news articles or expert networks to gauge IPO performance. These methods have yielded very inconsistent results and thus have left much to be desired. We attempt to alleviate the risk that an investor faces by introducing a framework for using sentiment analysis to forecast the first 3, 5, 10, 20, and 30 days price movement of an IPO. We illustrate that a model trained on an IPO prospectus's (Form S-1) sentiment can predict whether an IPO's stock price would increase or decrease from opening day. The prospectus is a formal document filed to the SEC that details the investment offering to the public as well as provides an important inside look into the IPO. Our research shows that a model trained on the sentiment of the prospectus can predict IPO price movement up to 9.6% higher than chance.