Hedging with Deep Reinforcement Learning and Deep Learning

Implementation of two deep reinforcement learning algorithms from

Hedging using reinforcement learning: Contextual k-Armed Bandit versus Q-learning
Loris Cannelli, Giuseppe Nuti, Marzio Sala, Oleg Szehr

and

Dynamic Replication and Hedging: A Reinforcement Learning Approach
P. N. Kolm and G. Ritter

We will also provide an approach based on classical deep learning, wich does not rely on reinforcement lerning.

The code provided is written in Python 3.8, and relies on the following libraries:

• torch
• numpy
• machin
• space-bandits