HIGH-FREQUENCY TRADING IMPLEMENTATION WITH STOCK PRICE PREDICTION

• Main Author: RASYADI PUTRAUTAMA (NIM : 13513046), BAYU
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/21398
• Institution: Institut Teknologi Bandung

High-Frequency Trading (HFT) is a form of algorithmic trading which utilizes computer speed to perform large amount of orders within 1 second. Institutions that use HFT don’t publicly provide documentations of their program. Documented simple HFT that has been implemented has a Sharpe Ratio below 1 which needs to be developed further and one way to do that is by predicting stock price. The main purpose of this final assignment is creating a prediction method that could improve the performance of HFT and also creating a HFT with a Sharpe Ratio above 1. <br /> <br /> <br /> Research started with exploring prediction methods which will be used, then the collected data analyzed followed by literature study to develop the solution’s designs. Each proposed solutions was then tested to obtain the most optimal solution for the system. <br /> <br /> <br /> Three prediction methods was selected to be used in the simulation, which are predictions using machine learning, support vector machine (SVM) and artificial neural network (ANN), and prediction using pattern matching which are used to determine the best time to buy or sell. Results from simulations with predictions shows better performance compared to simulations without predictions, especially the combination of SVM and pattern matching. Regardless, the technical indicators which are chosen for the simple HFT implementation are still inadequate.