THE USAGE OF CONVOLUTIONAL NEURAL NETWORK TO PREDICT STOCK PRICE MOVEMENT

• Main Author: Suparjo
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/47702
• Institution: Institut Teknologi Bandung

Stock is a booking unit that indicates the ownership of a certain company. The stock price which is fluctuating because of supply and demand makes many stock traders develop a way to gain maximum profit from the stock transaction. To gain profit, a deep analysis is needed in terms of the company factors and also historical stock price data. Along with time, Artificial Intelligence is mostly used in our daily life to make our life easier. Artificial intelligence can also be implemented in stock price analysis to identify the pattern of the historical stock price presented in a candlestick chart. One of the most popular ways to identify an object in picture format is the convolutional neural network. The convolutional neural network can be used to extract the features from a picture and then use the features to give an output. The output of the model will then be used to give an action recommendation of stock trading, as buy, sell, and hold. The convolutional neural network model that will be used for the experiment will be variated to get the best accuracy. The variables from time, like the interval to predict and input, will also be variated. The best model and input for every stock are dominated by a certain variation. The accuracy of the model can reach 73.3% with all of the total assets from the simulation with the model outperform the simulation by investing, buy, and hold until a certain time.