FORECAST OF STOCK TRENDS USING GAUSSIAN PROCESS MODEL

• Main Author: Wahyudi Handoko, Deaven
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/68975
• Institution: Institut Teknologi Bandung

In an era that is fully digitalized like today, the capital market has become one of the main choices of investing for the society. The stock market is one of the main choice for investing. A stock can be defined as a proof of ownership of a company, in which it will be sold at a price. The price of these stocks represents a company asset and the performance of the company, in which a company with a huge amount of asset and a good level of performance will have a higher price value, and vice versa. When the stock market is open, the prices of these stocks will move, following the rule of supply and demand. When the demand is higher than the supply, the stock price will rise, and vice versa. Investing in stocks is high risk, high reward. To minimize loss, statistical analysis can be used to forecast the price. With forecasting, a prediction of the stock price can be given using past values. In this study, the main method is gaussian process. This method can give a prediction with uncertainty. Gaussian process will be compared with SARIMA model. In this paper, gaussian process will use two covariance functions, matern and RBF. The input data will be split into three, short, medium, and long. This is done as each year give a different trend if compared with one another. By applying the model to two stocks, ASII and INDF, it can be concluded that gaussian process with matern covariance function gives a better model compared to SARIMA and gaussian process with RBF covariance function.