DOES THE USE OF ALGORITHMIC TRADING AIDED BY MACHINE LEARNING PROVIDE A CONSISTENT RETURN ON INVESTMENT?

• Main Author: Arya Wardhana H, Bimasena
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/86032
• Institution: Institut Teknologi Bandung

Investment is a process of allocating funds that aims to obtain profits in the long and short term. The use of algorithms in stock trading has been very popular lately because it can increase investment returns and reduce risk. The main objectives of this study are to identify criteria for assessing the consistency of investment returns, analyze the historical performance of trading algorithms, compare results with conventional strategies, determine the factors that affect the consistency of investment returns, and evaluate the risks associated with algorithmic trading. The data used includes the collection of historical data on BCA's share price in a certain period. In this study, the method used is linear regression and statistical analysis is carried out using the metrics of Mean Squared Error (MSE), Root Mean Squared Error (RMSE), accuracy, and Sharpe Ratio to evaluate the performance of the model. Model optimization is done through parameter tuning and feature selection, while cross- validation is used to avoid overfitting. Continuous monitoring and evaluation are carried out to ensure that the model remains effective. The results show that the trading algorithm using linear regression is able to produce a consistent return on investment with a low MSE value (close to zero) and high accuracy. The Sharpe ratio also shows that the trading model has a favorable return relative to the risk taken. Comparison with conventional trading strategies shows that linear regression methods have better return on investment and performance stability. Factors such as the type of strategy, market conditions, and algorithm parameters have been shown to have a significant effect on the consistency of investment returns. The conclusion of the study is that algorithmic trading powered by linear regression methods can provide more consistent and higher returns on investment compared to conventional trading strategies. However, it is important to continue monitoring, evaluation, and optimization of the model to manage risk and ensure optimal performance in the long term.