Forecasting Nifty Using Autoregressive Integrated Moving Average Model

Abstract

The study focuses on the forecasting capability of Auto Regressive Integrated Moving Average model for the nifty 50 index and how far the econometric models are dependable to prognosticate stock market indices. For this purpose, data have been collected from 15 June 2020 to 28 Jan 2022 Based on these data, forecasting is made from 18 Jan2022 to 28 Jan 2022. The series is converted into 1st difference as level data is not stationary, After making it stationary for nifty the researcher determines AR (p) lag and MA (q) lag through ACF and PACF tests. The ARIMA (1, 1, 3) model is accepted as the model that fulfills parsimony as this model having maximum significant variable, sigma square is minimum and Adjusted r 2 is maximum and AIC (Akaike info criterion) and SIC (Schwarz criterion) are minimum. The E-views 10 software package is used for ARIMA analysis. At last, it is seen that the ARIMA is not capable of forecasting data for a long period of time as time progresses it tends to forecast inaccurately. The investor cannot solely depend upon this model for their investment in securities market. Rather they must use other technical analysis tools.