OPAL: An Ontology-Based Framework with LSTM and PPO for Return Forecasting and Portfolio Optimization

• Igor Felipe Carboni Battazza UPE / FITec
• Cleyton Mário de Oliveira Rodrigues UPE
• João Fausto Lorenzato de Oliveira UPE

DOI: https://doi.org/10.5753/eniac.2025.11791

Resumo

Traditional portfolio optimization struggles with market volatility and nonlinear dynamics. We propose OPAL, a hybrid framework integrating semantic reasoning and machine learning for financial decision-making. OPAL combines ontology-based asset selection, Markowitz diversification, LSTM return forecasting, and PPO-based allocation. Using S&P 500 data (2015–2023), OPAL achieves a 118.05% cumulative return, 2.58 Sharpe ratio, and 4.99 Sortino ratio, surpassing baselines. Its modular design ensures transparency and adaptability.

Referências

Albadra, M. and Iqbal, R. (2019). An ontology-based model for credit scoring in peer-to-peer lending. Procedia Computer Science, 164:520–527.

Bhuyan, B. P. and Sastry, H. (2023). Stock selection using ontological financial analysis. International Journal of Information Technology and Management.

Buitelaar, P., Eigner, T., and Declerck, T. (2005). Ontology-based information extraction with soba. In LREC.

Fischer, T. and Krauss, C. (2018). Deep learning with long short-term memory networks for financial market predictions. European Journal of Operational Research, 270(2):654–669.

Hochreiter, S. and Schmidhuber, J. (1997). Long short-term memory. In Neural Computation, volume 9, pages 1735–1780.

Jiang, Z., Xu, D., and Liang, J. (2017). A deep reinforcement learning framework for the financial portfolio management problem. arXiv preprint arXiv:1706.10059.

Lillicrap, T. e. a. (2015). Continuous control with deep reinforcement learning. arXiv preprint arXiv:1509.02971.

Liu, Y., Li, Y., and Sun, Y. (2022). Forecasting the market with machine learning algorithms: An application of nmc-bert-lstm-dqn-x algorithm in quantitative trading. Mathematics, 10(3):347.

Markowitz, H. (1952). Portfolio selection. The Journal of Finance, 7(1):77–91.

Mnih, V. e. a. (2015). Human-level control through deep reinforcement learning. Nature, 518:529–533.

Mnih, V. e. a. (2016). Asynchronous methods for deep reinforcement learning. ICML.

Moody, J. and Saffell, M. (1998). Reinforcement learning for trading. Advances in Neural Information Processing Systems, 11:917–923.

Neuneier, R. (1998). Enhancing q-learning for optimal asset allocation. In NIPS.

Qin, Y., Song, D., Chen, H., Cheng, W., Jiang, G., and Cottrell, G. (2017). A dual-stage attention-based recurrent neural network for time series prediction. In IJCAI.

Reis, J., Silva, D., and Martinho, D. (2017). An ontology-based approach to portfolio construction. Expert Systems with Applications, 72:391–401.

Schulman, J. e. a. (2017). Proximal policy optimization algorithms. In arXiv preprint arXiv:1707.06347.

Siami-Namini, S., Tavakoli, N., and Siami Namin, A. (2018). A comparison of arima and lstm in forecasting time series. 2018 17th IEEE International Conference on Machine Learning and Applications (ICMLA), pages 1394–1401.

Wu, H. e. a. (2020). Deep transformer models for time series forecasting: A survey. arXiv preprint arXiv:2009.08317.

Ye, W., Xu, Y., and Lin, J. (2020). Financial portfolio allocation with hierarchical reinforcement learning. IEEE Access, 8:123208–123218.

Yu, L., Zhang, H., Feng, Y., and Xu, X. (2019). Deep reinforcement learning for portfolio management. Complexity, 2019.