Predicting IPO initial returns and failures



Initial Public Offering is a public investment process to offer the shares of a new stock to public. Companies generally decide to list the stocks in the market to make money for the company by selling shares in the market. The initial return of IPO can be defined as the difference in the IPO price and offer price on day one of IPO trading. The goal of an IPO is raising capital for the company. An IPO which cannot raise the required capital is considered to be an unsuccessful IPO. Most of the research done in this field is mainly concentrated on the simple regression models such as linear regression, logistic regression and random forest techniques. Hence, we will try to introduce advance methods like deep learning and neural networks in our research. The profitability of an IPO is the best measure to decide if the IPO was successful or a failure. Thus, the goal of this research was to use the data for an IPO that contains various numerical data with respect to the IPO performance as well as the company related information for IPOs traded in the market from the year 1996 to 2018 to predict the profitability of an IPO. We used four different deep learning techniques viz. simple recurrent neural networks, convolution neural networks, long short term memory and multilayer perceptron to predict the profitability and concluded that the simple recurrent neural networks give the best output based on various metrics such as accuracy, precision and recall.


A basic understanding of Python programming. A basic understanding of neural networks.

Speaker Info:

Hi, I am Parinita Hadkar. I am a datascientist with 2.5 years of experience in the field of software development. I recently completed my MSc in Datascience from Liverpool John Moore's University. I hold a Post Graduate diploma from International Institute of Information Technology, Bangalore and this topic was basically the part of my research project in my masters curriculum.

Speaker Links:

LinkedIn :

Section: Data Science, Machine Learning and AI
Type: Talks
Target Audience: Beginner
Last Updated: