Equal-information evaluation of credit assignment in modular financial reinforcement learning
Preprint
- 27 October 2026
- preprint
- Published by Elsevier in SSRN Electronic Journal
Abstract
Credit assignment in modular financial reinforcement learning is difficult to evaluate because an attributed signal may improve a component loss without adding decision-relevant information or improving economic outcomes. We propose an evaluation framework combining equal-information reconstruction controls, fixed-utility comparisons and prespecified downstream guardrails. The framework is instantiated in an attribution-guided modular trading architecture and tested in three controlled studies covering update allocation, supervision-target representation and portfolio risk control. We find that residual-based allocation is exactly reproduced by an ordinary loss-budget control under the disclosed interface; direct shortfall supervision improves component prediction but fails decoded-outcome preservation; and component-wise critic supervision worsens risk-condition certainty equivalent and drawdown relative to scalar supervision under fixed utility. These results identify three distinct evaluation failures: redundant attribution signals, representation-induced prediction trade-offs, and misalignment between component supervision and downstream risk outcomes. A preserved historical engineering case further shows why module-level learning claims require run-level provenance. The contribution is a reusable methodology for evaluating credit assignment in expert systems operating under partial observability, delayed feedback and economic risk constraints.Keywords
This publication has 12 references indexed in Scilit:
- Beyond random walks: exploring the learnability threshold of AI agents in algorithmic marketsExpert Systems with Applications, 2026
- Optimizing market-making strategies: A multi-objective reinforcement learning approach with pareto frontsExpert Systems with Applications, 2025
- NHITS: Neural Hierarchical Interpolation for Time Series ForecastingProceedings of the AAAI Conference on Artificial Intelligence, 2023
- Temporal Fusion Transformers for interpretable multi-horizon time series forecastingInternational Journal of Forecasting, 2021
- Empirical Asset Pricing via Machine LearningThe Review of Financial Studies, 2020
- DeepLOB: Deep Convolutional Neural Networks for Limit Order BooksIEEE Transactions on Signal Processing, 2019
- The Price Impact of Order Book EventsJournal of Financial Econometrics, 2013
- Optimal execution of portfolio transactionsJournal of Risk, 2001
- A New Approach to the Economic Analysis of Nonstationary Time Series and the Business CycleEconometrica, 1989
- The Combination of ForecastsJournal of the Operational Research Society, 1969