ActorCritic_Binomial_Option_Hedging

One simple Python code to implement Actor Critic Algorithm for hedging European At-The-Money option under Binomial Tree model.

This is one of projects completed at HKUST MSc in Financial Mathematics (MAFM) program , 2021 Spring, for the course MAFS6010Y Reinforcement Learning with Applications in Finance.

Reward Design is closely related to the idea of hedge, NOT to make positive or negative PnL but to make your portfolio (one option + $\delta$ stocks) to have NO volatilities in value. Here I choose $-abs(R)$ as the reward for each steps of one episode.

Binomial Tree is the most simple option pricing techniques with emphasis on replication. One environment is created to represent this pricing idea. Environment is an important concept in reinforcement learning mainly to interact with the agent (i.e., the hedger) and offers state & reward update.