Abstract
Artificial neural networks are extensively used to predict the financial time series. This study implements the neural network model for predicting the daily returns of the Pakistan Stock Exchange (PSE). Such an application for PSE is very rare. A multi-layer perception network is used for the model used in this study, while the network is trained using the Error Back Propagation algorithm. The results showed that the predictive power of the network was performed by the return of the previous day rather than the input of the first three days. Therefore, this study showed satisfactory results for PSE. In short, artificial intelligence can be used to give a better picture of stock market operators and can be used as an alternative or additional to predict financial variables.
Highlights
Predictability of stock returns always remained vital for computational scientists as this involves enormous amount of monetary benefits
Pakistan Stock Exchange (PSE) is a developing equity market and we found very few studies in the field by using artificial neural network and those which are being found used smaller sample size as compare to the current study the architecture of this study is comprehensive, which is discussed in the latter sections
This article is an attempt to predict the daily return on the stock market for PSE
Summary
Predictability of stock returns always remained vital for computational scientists as this involves enormous amount of monetary benefits. Whereas Neural Networks appeared to be very helpful in predicting non-liner trends and volatilities of time series which can be considered one of the major contribution of the artificial intelligence in recent years. These models can be used to analyze the relations between economic and financial variables, data filtration, forecasting, optimization and generating time series (Cameron & Scuse, 1999; Cao et al, 2005; Cheh, 190 Vol 4, Issue 2 ISSN 2414-2336 (Print), ISSN 2523-2525 (Online). Neural networks have got significant reception by the financial engineers and practitioners in recent years because of their immense learning abilities Supporters of these models include academicians, practitioners and industry persons like researchers, portfolio managers, investment banks, trading firms and most of the major investment banks. Neural networks application in finance include risk measurement for the mortgage loans (Collins, Gohsh, & Scofield, 1988), corporate bonds are being rated by using neural network framework (Altman, et al, 1994; Salchenberger, Cinar, & Lash, 1992), credit cards rating (Susan & Chye, 1997), pricing of derivatives (Hutchinson, 1994) so on and so forth
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
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
