Introduction to Financial Deep Learning
Often times we have an optimal, diversified portfolio, (maybe created by using my pipeline) and we want to manage the portfolio according to a predetermined set of rules. Instead of building an algorithmic trading system with investment management based on control we can build a neural network to act on our portfolio by training it on certain instructions. In this article, we will be using Python and its accompanying libraries to build this neural network. To give you an idea of how this can be implemented, consider a diversified portfolio in which we wish to mitigate systemic risk. We can train a neural network to liquidate a portfolio if it declines by x% any given day acting like a personal circuit breaker. Obviously, that condition is arbitrary and the model can be trained to follow any annotated signal you specify.
Originally Posted by Roman Paolucci | Artificial Intelligence on Medium| Startup| February 7th 2020