Summer Research Fellowship Programme of India's Science Academies

Application of time series forecasting in finance and business

Kaushik Ramnath G

SRM Institute of Science and Technology, Kattankulathur, Tamil Nadu 603203

K. S. Mallikarjuna Rao

Associate Professor, Industrial Engineering and Operations Research, IIT Bombay, Mumbai Maharasthra


Predicting the future is one of the important problems in practical application especially in industries like finance and business. The prevailing theory is that stock prices are random and unpredictable. In recent years, computer science combined with advanced mathematics is revolutionizing the world of finance and business. Time series forecasting is one such field of study in which predictions can be made with complete mathematical proofs. In finance, using these predictions computers can be automated using programming languages like R/Python for triggering buy and sell signals. Many factors such as economic conditions, political events, and traders’ expectations may have an influence on the stock market index. Predicting stock trends can minimize the risk of investing and maximize profit. Mathematical concepts like linear algebra, statistics, and probability play a huge role in my work. My work discusses different types of time series models and its mathematical explanations for predicting the future. Time series models like ARIMA, Exponential Smoothing, Holt-Winters are some of the popular models for predictions. A Deep Learning model known as LSTM is also explained and implemented using Python. My work tries to predict the price movements of NIFTY50 index values using mathematical time series models which are implemented in programming languages like Python/R. It is then used to formulate a strategy to bet on future stock prices for making a profit. A type of accuracy measures called RMSE is used for validating the models used in datasets. I intend to obtain results that reveal that the time series modeling with the complete mathematical understanding and programming knowledge has a strong potential for short-term prediction of not only stock market trends but also business-related trends.

Keywords: ARIMA, NIFTY50, index, Holt-Winters, Deep Learning

Kaushik Ramnath G

Written, reviewed, revised, proofed and published with