Abstract
An initial public offering (IPO) is a type of public offering in which a company’s shares are sold to institutional and individual investors. While the majority of studies on IPOs have focused on the efficiency of raising capital and price adequacy in IPOs, studies on portfolio allocation strategies for IPO stocks are relatively scarce. This paper develops a machine learning investment strategy for IPO stocks based on rough set theory and a genetic algorithm (GA-rough set theory). To reduce issues of information asymmetry, we use nonfinancial data that are publicly available to individual and institutional investors in the IPO process. Based on the rule sets generated from the training sets, we conduct 120 tests with various conditions involving the target days and the partition of the training and testing sets, and we find excess returns of the constructed portfolios compared to the benchmark portfolios. Investors in IPO stocks can formulate more efficient investment strategies using our system. In this sense, the system developed in this paper contributes to the efficiency of financial markets and helps achieve sustained economic growth.
Highlights
An initial public offering (IPO) refers to the process of offering new shares in a private corporation to the public in a new stock issuance
This study presents a machine learning investment strategy to maximize the returns on investment in IPO stocks, and it identifies the degree of improvement in the returns by comparing the performance of the machine learning strategy with benchmarks
We use nonfinancial data disclosed in the IPO process and employ machine learning portfolio allocation method that does not require estimating input parameters from market data
Summary
An initial public offering (IPO) refers to the process of offering new shares in a private corporation to the public in a new stock issuance. In the IPO process, individual investors cannot take advantage of technical analyses of stock prices and trading volume or fundamental analyses that utilize financial indicators for investments in listed stocks. We use nonfinancial data disclosed in the IPO process and employ machine learning portfolio allocation method that does not require estimating input parameters from market data. Investors in IPO stock markets are able to create more efficient portfolios using our machine learning portfolio allocation system based on GA-rough set theory. In this sense, the system developed in this paper appears to contribute to the efficiency of financial markets, and it plays a role in sustaining economic growth. The experimental results and the analysis of the results are discussed in Section 4, and Section 5 offers concluding remarks
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
