The focus of my research is the implementation of the Avellaneda Stoikov market making model, which is managed by Reinforcement Learning (RL). The RL is designed to dynamically configure the various parameters of the Avellaneda Stoikov model, taking into account both market and private states.
The Avellaneda Stoikov market making model, when used in conjunction with Reinforcement Learning (RL), offers several advantages:
- Dynamic Parameter Configuration: RL can dynamically adjust the parameters of the Avellaneda Stoikov model based on the current market and private states. This allows the model to adapt to changing market conditions, potentially improving its performance.
- Risk Management: The Avellaneda Stoikov model is designed to minimize inventory risk, which is a significant concern in market making. By using RL to adjust the model's parameters, it's possible to further optimize this risk management.
- Learning from Experience: RL algorithms learn from their past actions and the resulting rewards or penalties. This means that over time, the RL-enhanced Avellaneda Stoikov model can improve its performance by learning from its past trades.
- Potential for Improved Performance: The combination of the Avellaneda Stoikov model with RL has the potential to outperform traditional market making strategies, as demonstrated in academic research.
- Flexibility: The RL can be trained with different types of algorithms (like DQN, PPO, etc.), providing flexibility in
choosing the most suitable one for the specific market conditions and trading objectives.