Stock Price Prediction and Portfolio Optimization Using Hidden Markov Model and Quadratic Programming

This study introduces a decision support system for optimizing stock portfolios. It employs a Hidden Markov Model for 95% accurate stock price forecasts over a five-day period. The system utilizes Quadratic Programming to refine allocation, focusing on up to 15 blue-chip companies. Key indicators like MACD, RSI, and Bollinger Bands offer clear buy, sell, or hold recommendations. It actively monitors the Philippine Stock Exchange, notifying users of significant deviations. Developed in Python and C#, both models were trained on a consistent 5-year dataset for reliable performance. Results show high prediction accuracy, averaging 98.923% (high price) and 98.741% (low price), benefiting investment strategies. The study compared SPOT-allocated portfolios to random ones, revealing a notable difference in profit/loss allocation. This underscores its value for informed and profitable decision-making in stock trading.