DEVELOPMENT OF SOFTWARE TO PREDICT STOCK PRICE USING AUTOREGRESSIVE INTEGRATED MOVING AVERAGE (ARIMA) MODEL

• Main Author: Benigno Letupeirissa (NIM : 13501020), Adrian
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/7045
• Institution: Institut Teknologi Bandung

Abstract: <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> There is a law that is generally known in investment. The law states that an investment media which has higher possibility of profit will be followed by higher risk. The risk takes form in the potential of failure in making prediction of future value or price. Predicting stock price is not easy because of uncertain fluctuation in stock market. Thus an investor should analyze stock market data and external factors in order to avoid big loss. The analyzed stock data are time-series data. A time-series is defined as a record taken trough time, that is a sequential set of data measured over time. If those data are modeled, then it can be shown that they have a trend. By modeling data trend, the value of next observation can be predicted. A method that can be used to model time-series data trend is Autoregressive Integrated Moving Average (ARIMA) model. ARIMA is a model based on Linear Regression concept. ARIMA model, which is produced by historical analysis to stock time-series data, can be used to predict future stock price. This final project is focused on prediction of stock price using Autoregressive Integrated Moving Average model, which consist of procedure of ARIMA modelling for a set of data, searching adequate model for the data, and predicting future data value. In this final project, a software named SeFA (Stock Forecast with ARIMA) is also built to implement ARIMA model in modelling and predicting future stock price. SeFA is developed using Borland Delphi 7.0 in Windows Operating System environment. Stock data used by SeFA as input are in text file (.txt) with specified writing format. SeFA is then used to look for an ARIMA model that is adequate for common stock data, data trend, and accuracy level of prediction. Testing process is done by comparing prediction result for 20 last days of the input and the actual data. Testing result shows that ARIMA(2,1,1) is the adequate model for common stock data. Testing result also shows that trend modelling that is resulted by SeFA has a high level of accuracy, while in the same time also has a low level of accuracy in prediction.