Deep Neural Network-Based Financial Forecasting Through Automated Feature Extraction and Adaptive Model Optimization
Keywords:
financial forecasting; deep neural networks; automated feature extraction; adaptive optimization; nonstationarity; uncertainty calibrationAbstract
Financial forecasting with deep neural networks is constrained by two coupled design problems: informative features must be constructed from heterogeneous, noisy, and asynchronously released signals, and model parameters must remain effective as market conditions change. Manual feature engineering is difficult to maintain across assets and horizons, while static hyperparameter choices often become poorly matched to evolving regimes. This manuscript develops an integrated methodological framework for financial forecasting that couples automated feature extraction with adaptive model optimization. The representation component combines variable-specific encoders, temporal recurrence or convolution, and attention-based aggregation with group-sparse regularization, so that predictive features are learned from raw and moderately processed inputs while irrelevant variable blocks can be suppressed. The optimization component separates inner parameter estimation from outer hyperparameter and architecture adaptation through a bilevel objective, and it supports regularized online updates for controlled adaptation after initial training. Predictive uncertainty is treated as part of the forecasting object rather than an afterthought, using ensemble dispersion and residual variance estimates to inform calibration and risk-aware evaluation. The framework specifies leakage-resistant rolling-origin validation, purged temporal splits, and statistical comparison against credible baselines. Analytical assessment identifies the main trade-offs among representation flexibility, overfitting risk, adaptation speed, stability, interpretability, and computational cost. The contribution is a coherent specification of how automated representation learning and adaptive optimization can be combined and evaluated for financial forecasting without assuming that additional model complexity automatically improves predictive accuracy.
References
1. LeCun, Y., Bengio, Y., & Hinton, G. (2015). Deep learning. Nature, 521(7553), 436–444.
2. Goodfellow, I., Bengio, Y., & Courville, A. (2016). Deep learning. MIT Press.
3. Fischer, T., & Krauss, C. (2018). Deep learning with long short-term memory networks for financial market predictions. European Journal of Operational Research, 270(2), 654–669.
4. Gu, S., Kelly, B., & Xiu, D. (2020). Empirical asset pricing via machine learning. The Review of Financial Studies, 33(5), 2223–2273.
5. Luo, D., Chen, D., Wang, J., Zhu, G., & Xu, W. (2020). The smart robot crafting approach to computing materials. Construction Robotics, 4(3), 239-249.
6. Sezer, O. B., Gudelek, M. U., & Ozbayoglu, A. M. (2020). Financial time series forecasting with deep learning: A systematic literature review: 2005–2019. Applied Soft Computing, 90, 106181.
7. Hochreiter, S., & Schmidhuber, J. (1997). Long short-term memory. Neural Computation, 9(8), 1735–1780.
8. Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A. N., Kaiser, Ł., & Polosukhin, I. (2017). Attention is all you need. Advances in Neural Information Processing Systems, 30, 5998–6008.
9. Kingma, D. P., & Ba, J. (2015). Adam: A method for stochastic optimization. In Proceedings of the 3rd International Conference on Learning Representations.
10. Finn, C., Abbeel, P., & Levine, S. (2017). Model-agnostic meta-learning for fast adaptation of deep networks. In Proceedings of the 34th International Conference on Machine Learning (Vol. 70, pp. 1126–1135). PMLR.
11. Hospedales, T., Antoniou, A., Micaelli, P., & Storkey, A. (2022). Meta-learning in neural networks: A survey. IEEE Transactions on Pattern Analysis and Machine Intelligence, 44(9), 5149–5169.
12. Gama, J., Žliobaitė, I., Bifet, A., Pechenizkiy, M., & Bouchachia, A. (2014). A survey on concept drift adaptation. ACM Computing Surveys, 46(4), 1–37.
13. López de Prado, M. (2018). Advances in financial machine learning. Wiley.
14. Bergmeir, C., & Benítez, J. M. (2012). On the use of cross-validation for time series predictor evaluation. Information Sciences, 191, 192–213.
15. Lakshminarayanan, B., Pritzel, A., & Blundell, C. (2017). Simple and scalable predictive uncertainty estimation using deep ensembles. Advances in Neural Information Processing Systems, 30, 6402–6413.
16. Gal, Y., & Ghahramani, Z. (2016). Dropout as a Bayesian approximation: Representing model uncertainty in deep learning. In Proceedings of the 33rd International Conference on Machine Learning (Vol. 48, pp. 1050–1059). PMLR.
17. Diebold, F. X., & Mariano, R. S. (1995). Comparing predictive accuracy. Journal of Business & Economic Statistics, 13(3), 253–263.
18. Hansen, P. R., Lunde, A., & Nason, J. M. (2011). The model confidence set. Econometrica, 79(2), 453–497.
19. Makridakis, S., Spiliotis, E., & Assimakopoulos, V. (2018). The M4 Competition: Results, findings, conclusion and way forward. International Journal of Forecasting, 34(4), 802–808.
20. Lim, B., Arik, S. O., Loeff, N., & Pfister, T. (2021). Temporal fusion transformers for interpretable multi-horizon time series forecasting. International Journal of Forecasting, 37(4), 1748–1764.
Downloads
Published
How to Cite
Issue
Section
License
Copyright (c) 2026 Global Financial Analytics Research Review

This work is licensed under a Creative Commons Attribution 4.0 International License.
This article is published under the Creative Commons Attribution 4.0 International License (CC BY 4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.



