Guaranteed funds’ replication by reinforcement learning

Guaranteed funds provide an important protection vehicle for risk-averse investors during times of increasing market volatility. Protection may be on the capital invested or its return or both: the associated market risk is then transferred onto fund managers and financial institutions, facing, as a result, an optimal hedging problem. In this article we focus on capital protection and address the fund manager replication problem relying on a reinforcement learning (RL) approach. The article clarifies the financial and methodological implications of the RL method and a case study is presented relying on a 1-year guaranteed fund, whose replication by the issuer is tested over the 2018–2024 period. A US fund manager is assumed with a decision space including cash, bond and equity assets, and the optimal replication of the guaranteed fund is formulated considering a telescoping horizon leading to quarterly portfolio revisions. By combining the learning process and the derivation of the optimal policy in a recursive manner, the RL method is able to solve the minimum guarantee fixed-mix problem in minutes, significantly improving classical solution times in stochastic optimization, yet considering a very rich uncertainty model based on a data-driven simulation method and consistent with arbitrage-free conditions in the market.