Machine learning models have become increasingly popular in the financial markets because of their usability in the trading desks. This study explores the application of different machine learning algorithms to develop an automated trading system for the ETF ‘QQQ’ (which is a tracker of the NASDAQ-100 Index). Our aim is to build a model capable of consistently generating positive returns over time by predicting buy and sell signals that help the investor making investment decisions. The methodology involves using historical price data and technical indicators, particularly the Moving Average Convergence Divergence (MACD) as input features. We compare several machine learning techniques, including XGBoost, Random Forest, and Logistic Regression. Data preparation of our models is based on sequence splitting, feature scaling, and dimensionality reduction through Principal Component Analysis (PCA). The models are trained and evaluated using cross-validation and hyper-parameter tuning via Grid Search. Our results indicate that XGBoost shows the highest return of 34.13 percent over the test period. The Random Forest algorithm demonstrates the highest accuracy with an f1-score of 0.744. The outcome of our models show the appropriateness of Machine Learning algorithms when applied to testing strategies due to their predictive ability and enhancement capabilities.