kalyan678
commented
1 year ago Hey @YorkLouis - The optimization framework used in deep learning for portfolio optimization depends on the specific approach taken. If using neural networks to predict asset returns, any traditional optimization framework can be used. If using deep reinforcement learning to directly learn optimal portfolio weights, the agent learns to take actions to maximize a reward signal. The main difference between supervised learning and deep reinforcement learning is that the former involves learning a mapping between input and output labels, while the latter involves learning to make sequential decisions that maximize a reward signal. Supervised learning can be used for predicting asset returns, while deep reinforcement learning can be used to directly learn optimal portfolio weights. As far as I know, portfolio optimization is still a relatively new and active area of research, and there are many challenges and limitations that need to be addressed. Additionally, the use of deep learning in finance requires careful consideration of ethical and regulatory issues.
Below are the some of the research papers that you can explore for even better understanding :
- A Deep Reinforcement Learning Framework for the Financial Portfolio Management Problem" by Zhengyao Jiang, Dixing Xu, and Jinjun Liang (2017) - https://arxiv.org/abs/1706.10059
- Deep Learning in Asset Pricing" by Bryan Kelly, Yuncheng Li, and Jiamin Xu (2019) - https://arxiv.org/pdf/1904.00745.pdf
- Portfolio selection using neural networks by Alberto Fernández, Sergio Gómez - https://arxiv.org/ftp/cs/papers/0501/0501005.pdf
Hope this information helps!
pepi99
What optimization framework can this method be applied? What the difference with supervised learning, and Deep Reinforcement Learning?