FORECASTING STOCK PRICES IN PROPERTY DEVELOPMENT GROUP IN THE STOCK EXCHANGE OF THAILAND BASED ON STATISTICAL MODELS

Abstract

The objective of the study regarding to Forecast Stock Price index of Property and Construction Sector on the Stock Exchange of Thailand with more than 10 new project plans launched in 2023 for a period of 16 years from January 1, 2007 to December 31, 2022 of 10 securities namely AP, SIRI, SPALI, LH, FPT, LALIN, NOBLE, PF, LPN and SC The time-series techniques consisted of the Box-Jenkins method, and the exponential smoothing through seasonal additive and multiplicative models to forecast stock prices.

The results of the study forecasting stock prices using secondary data in terms of monthly time series showed that 1) the ARIMA method was the most appropriate to forecast for 5 variables consist of LH, FPT, LALIN, PF and LPN as it gave the lowest value of root mean square error equal (0.311), (0.331), (0.380), (0.005) and (0.179) respectively, 2) SARIMA method was the most appropriate to forecast for 5 variables consist of SIRI, LH, FPT, PF and LPN as it gave the lowest value of root mean square error equal (0.092), (0.311), (0.331), (0.005) and (0.179) respectively, 3) the Seasonal Additive Exponential Smoothing Technique was the most appropriate to forecast for 2 variables consist of AP and SPALI as it gave the lowest value of root mean square error equal (0.582) and (1.256) respectively, and 4) the Seasonal Multiplicative Exponential Smoothing Technique was the most appropriate to forecast for 2 variables consist of NOBLE and SC as it gave the lowest value of root mean square error equal (0.483) and (0.235) respectively. Therefore in this study, SARIMA(p,d,q) model was the most appropriate for time series forecasting which is very accurate in both short and medium term forecasting.