ARIMA modeling: Forecasting Philippine oil company stock prices

• Main Authors: Gonzales, Ervin John, Munar, Philipp Andrew., San Luis, John Paul, San Luis, Wilfrid
• Format: text
• Language: English
• Published: Animo Repository 2017
• Subjects: Stocks > Prices > Philippines; Petroleum products > Prices > Philippines; Finance and Financial Management
• Online Access: https://animorepository.dlsu.edu.ph/etd_bachelors/9051
• Institution: De La Salle University

This paper examined time, trends, seasonalities, and cycles to attempt to forecast stock price directionality utilizing the Box-Jenkins models. The research methodology involved collecting past stock prices to statistically regress fitted models which were used to project stock price forecasts. Furthermore, this paper defined the viability of autoregressive integrated moving average (ARIMA) models, also known as Box-Jenkins models, in dealing with time series data dor the purpose of forecasting. This study particularly on Philippine publicly-listed companies in the Oil sector and the subsectors that fall under it. Sample data used for the chosen companies observed a weekly interval, and spanned from January 2006 to December 2015. The models for each company dataset were sebjected to various tests such as Akaike Information Criterion, Bayesian Information Criterion, and autocorrelation functions to guarantee the most parsimonious model. In the accuracy testing phase, mean absolute percentage error (MAPE) was used to quantify how precise the stock price predictions from our final models were for January 2016 to October 2017. This paper's results showed significant accuracy in forecasting stock prices using ARIMA models. The findings indicate that ARIMA is appropriate for handling time series data, and it can be concluded its robust forecast capability is applicable to stock price prediction.