Aplicações de inteligência artificial e reinforcement learning na previsão de preços e gestão de portfólios: ensaios no contexto financeiro e de commodities

Abstract

This thesis comprises three essays that utilize artificial intelligence (AI) and machine learning for price prediction and portfolio optimization. The first essay applies LSTM neural networks to predict Brazilian ethanol prices, demonstrating that this model outperforms other algorithms in accuracy, especially during the pandemic. The second essay conducts a systematic literature review on AI in portfolio management, identifying gaps and suggesting future directions. The third essay compares reinforcement learning algorithms with traditional strategies, showing that the inclusion of derivatives enhances portfolio performance and that RL techniques increase returns without raising risks. The results highlight the effectiveness of AI in price predictions and investment optimization, offering advantages in the financial and agricultural commodities sectors.