Stock market prediction using time series analysis

Abstract

Stock market, a very unpredictable sector of finance, involves a large number of investors, buyers and sellers. Stock prediction has been a phenomenon since machine learning was introduced. But very few techniques became useful for forecasting the stock market as it changes with the passage of time. As time is playing a crucial rule here, Time Series (TS) analysis is used in this paper to predict short-term stock market. The first step for analyzing TS is to check whether historical stock market data is stationary using Plotting Rolling Statistics and Dickey-Fuller Test. Secondly, Trend and Seasonality is eliminated from the series to make the data a stationary series. Then, TS stochastic model known as Autoregressive Integrated Moving Average (ARIMA) is used as it has been broadly applied in financial and economic sectors for its efficiency and great potentiality for short-term stock market prediction. For comparing the performance, the three subclasses of ARIMA such as: Autoregressive (AR), Moving Average (MA), and Autoregressive Moving Average (ARMA) are also applied. Finally, the forecasted values are converted to the original scale by applying Trend and Seasonality constraints back. KEYWORDS: Stock Prediction, Machine Learning, Time Series, ARMA, ARIMA.