ON EFFICIENCY OF AUTOREGRESSIVE MODELS IN BSE FINANCE SENSEX WITH FORECASTABILITY ASSESSMENT

Abstract

The Indian stock market's finance sector is incredibly active and erratic. It's never easy to forecast the finance sector.  The current study aims to develop forecasting models for the sensex of the finance sector. The value of a sector-specific model for forecasting stock prices or returns within that specific industry is then assessed. Sector-specific forecasting models also aim to be compared with other sector-wise models, and more precisely with stock price forecasting models, by comparing relevant predicting errors that each model generates. So, the main objective is to ascertain the viability of sector-wise forecasting models for stock price or return forecasting with an acceptable approximation error. Based on daily sensex data, sector-specific sensex values are predicted using the Box-Jenkins autoregressive integrated moving average (ARIMA). The data must be differenced in order to make an integrated time series stationary. Lags of the forecast errors and lags of the values of the original variable are used as regressors in the ARIMA model. Once the best model has been identified, one candidate stock from each segment and the Finance sector is chosen. The suitability of the sector and segment model for the chosen stock price projections is looked at. The forecasting error is then compared to the error value of the stock price forecasting model. The gap dictates whether the sector- and segment-wise models apply to the corresponding equities. Lastly, sector & segment models are applied to five large cap companies from the Finance sector, which are chosen in order to cross-validate the forecasting models' applicability. The validity of the sensex forecasting models unique to sectors and segments is further investigated by comparing the error numbers for each of the five equities in relation to stock price predictions. The forecasting model used in this study for the finance industry has minimal prediction errors over a variety of ARIMA parameter values. Investors might use SE models when making investments in different markets or businesses.  Assets or funds can be allocated by a portfolio manager or investor according to their investment strategy to different sectors or segments. A stock that belongs to a certain industry and market segment can have its performance and price predicted utilizing linked intersecting forecasting models.

Keywords: ARIMA, Finance, Forecasting, Random Walk, Performance, Sensex.