Neurosymbolic AI reduces portfolio rule violations in continuous allocation
Internalizing Investment Mandates: A Neurosymbolic RL Approach to Portfolio Allocation
Machine Learning
Summary
Investment portfolios often have rules that say how money should be spread across different assets, depending on market conditions. The authors show a way to train AI systems that not only try to make good financial decisions but also follow these specific rules without needing extra checks later. Their method uses a mix of neural networks and symbolic logic to learn rule compliance directly while deciding how to invest. This approach cuts down mistakes in following rules from nearly all cases to under 5%, making AI more reliable for managing portfolio allocations.
What this means in practice
- •For financial portfolio managers: Create AI-driven investment strategies that respect detailed allocation rules without manual intervention during trading.
- •For algorithmic trading teams: Develop trading algorithms that internally enforce complex regulatory or policy constraints while optimizing returns in continuous action settings.
Authors
Zeynep Yıldız, Satchit Chatterji, Ercan Atam, Erman Acar
Abstract
Institutional portfolios operate under investment policy statements that specify asset allocation constraints, often conditioned on market regime information. Deep reinforcement learning has been shown to optimize financial performance but typically does not enforce rule compliance, making it unsuitable for real-world deployment. Existing neurosymbolic methods such as Probabilistic Logic Policy Gradient (PLPG) offer constraint-satisfying reinforcement learning (RL) via differentiable symbolic supervision but assume a finite, discrete action space. We present a PLPG-inspired continuous allocation method that estimates the probability that the policy's own executed action satisfies the investment mandate, using reparameterized Monte Carlo sampling, and adds it to the training objective. No post-hoc enforcement mechanism intercepts actions at inference time and compliance is learned by the agent. Audited directly on deterministic mean actions with no correction, the policy reduces violations of a six-rule regime-conditional mandate from 96-98% for unconstrained RL and 33.4% for a reward-penalty baseline to 4.8%, showing that a continuous-action neurosymbolic policy can substantially internalize a symbolic mandate.