Snapshots of Model-Free Deep Hedging with Transaction Costs and Light and Augmented Data Methods

Key Points

• The aim is to explore the effectiveness of model-free deep hedging networks in small datasets, focusing on their performance against traditional hedging methods.
• Developed a lightweight, model-free deep hedging network.
• Conducted tests using geometric Brownian motion with high transaction costs.
• Evaluated performance against traditional hedging methods like Black–Scholes and Leland.
• The model-free deep hedging network outperformed Black–Scholes and Leland hedges in tests.
• Effective hedging was achieved even with limited data, enhancing usability in investment processes.