Brandon Pennington
Quantitative Finance Researcher
Imperial College London | London, UK
About
A quantitative finance researcher at Imperial College London focusing on the application of reinforcement learning to real estate markets and derivative pricing.
I am interested in how RL agents can learn optimal strategies in illiquid markets with complex dynamics, bridging financial modeling with machine learning techniques. My research builds on foundational work in stochastic control for finance and recent advances in deep reinforcement learning, with particular emphasis on partially observable environments, long-horizon credit assignment, and robust policy learning under model misspecification.
Research Interests
- Reinforcement learning & simulation
- Stochastic interest rate models (Hull-White, Vasicek, CIR)
- Market simulation with illiquidity modeling
- Portfolio optimization under transaction costs and constraints
- Regime-switching models and volatility clustering in markets
Selected Reading & Influences
- Buehler et al. (2019) – “Deep Hedging” (Journal of Financial Economics)
- Jiang et al. (2017) – “A Deep Reinforcement Learning Framework for the Financial Portfolio Management Problem”
- Cao et al. (2021) – “Real Estate Option Pricing with Illiquidity and Stochastic Interest Rates”
- GARLEANU and Pedersen (2013) – “Dynamic Trading with Predictable Returns and Transaction Costs” (Journal of Finance)
- Hull and White (1990, 1994) – Term structure models and extensions
- Dixon and Chong (2023) – “Real Estate Derivatives: A Reinforcement Learning Approach” (working paper)
- Fleming and Riseth (2023) – “Model-Free Reinforcement Learning for Option Hedging”
Current Project
re_gym
A Gymnasium-compatible reinforcement learning environment for training agents on real estate option trading and derivative strategies with realistic market dynamics.
The library models real estate market characteristics including illiquidity, transaction costs, bid-ask spreads, and mean-reverting property values with regime-switching volatility.
Environments
- LeaseOption-v1 – Single property with exercise decisions
- PropertyPortfolio-v1 – Multi-property acquisition and disposition
- MortgageHedging-v1 – Interest rate hedging strategies
- REITTrading-v1 – REIT allocation with property simulation
Features
- Stochastic interest rate models (Hull-White, Vasicek, CIR)
- Compatible with Stable-Baselines3 and FinRL
- UK Land Registry and FRED data integration
Contact
Email: BrandPennington@protonmail.com