Forecasting Liquor Index: Applications of ARIMA and Gray Model

Abstract

An imperfect stock market provides ambitious investors with plenty of room for arbitrage. Currently, ARIMA (Autoregressive Integrated Moving Average) and Gray forecast models are widely used to forecast future stock prices. In our paper, we aim to investigate the efficiency of those two models. Since each stock is idiosyncratic by nature, it is not advisable to look for one single forecast method that is robust for forecasting all the stocks in the market. Thus, we select the Chinese liquor industry, one of the most popular industry to invest in for the past year in China, as the context to study these two models. To compare the efficiency of those two models for forecasting the Chinese liquor index specifically, we first format the data into the appropriate form and then test various assumptions hidden in ARIMA and Gray forecasting models. After applying ARIMA and Gray forecast models respectively to predict stock closing prices for 20 days, we conclude that the ARIMA model performs better with a smaller sum of squared residuals. As a result, we deduce that ARIMA has a better ability of forecasting Liquor Index than Gray model. Methods incorporating more factors to forecast index deserve further investigation.