{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,10,27]],"date-time":"2026-10-27T14:37:27Z","timestamp":1793111847555,"version":"4.5.0"},"posted":{"date-parts":[[2026]]},"group-title":"SSRN","reference-count":37,"publisher":"Elsevier BV","license":[{"start":{"date-parts":[[2026,1,1]],"date-time":"2026-01-01T00:00:00Z","timestamp":1767225600000},"content-version":"unspecified","delay-in-days":0,"URL":"https:\/\/www.uspto.gov\/ip-policy\/copyright-policy\/copyright-basics"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"abstract":"<jats:p>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.<\/jats:p>","DOI":"10.2139\/ssrn.7674869","type":"posted-content","created":{"date-parts":[[2026,10,27]],"date-time":"2026-10-27T14:00:22Z","timestamp":1793109622000},"source":"Crossref","is-referenced-by-count":0,"title":["Equal-information evaluation of credit assignment in modular financial reinforcement learning"],"prefix":"10.2139","author":[{"ORCID":"https:\/\/orcid.org\/0009-0005-6276-0958","authenticated-orcid":true,"given":"Gennadiy","family":"Chistiakov","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-4651-7585","authenticated-orcid":true,"given":"Yury","family":"Yanovich","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"78","reference":[{"key":"ref1","doi-asserted-by":"crossref","first-page":"5","DOI":"10.21314\/JOR.2001.041","article-title":"Optimal execution of portfolio transactions","volume":"3","author":"R Almgren","year":"2001","journal-title":"Journal of Risk"},{"key":"ref2","article-title":"Self-citation masked for double-blind review","author":"Anonymous","year":"2027","journal-title":"Expert Systems with Applications"},{"key":"ref3","doi-asserted-by":"crossref","first-page":"94","DOI":"10.3905\/jpm.2014.40.5.094","article-title":"The deflated sharpe ratio: Correcting for selection bias, backtest overfitting, and non-normality","volume":"40","author":"D H Bailey","year":"2014","journal-title":"The Journal of Portfolio Management"},{"key":"ref4","doi-asserted-by":"crossref","first-page":"451","DOI":"10.1057\/jors.1969.103","article-title":"The combination of forecasts","volume":"20","author":"J M Bates","year":"1969","journal-title":"Journal of the Operational Research Society"},{"key":"ref5","doi-asserted-by":"crossref","first-page":"6989","DOI":"10.1609\/aaai.v37i6.25854","article-title":"NHITS: Neural hierarchical interpolation for time series forecasting","author":"C Challu","year":"2023","journal-title":"Proceedings of the AAAI Conference on Artificial Intelligence"},{"key":"ref6","doi-asserted-by":"crossref","first-page":"47","DOI":"10.1093\/jjfinec\/nbt003","article-title":"The price impact of order book events","volume":"12","author":"R Cont","year":"2014","journal-title":"Journal of Financial Econometrics"},{"key":"ref7","first-page":"10148","article-title":"A decoder-only foundation model for time-series forecasting","author":"A Das","year":"2024","journal-title":"Proceedings of the 41st International Conference on Machine Learning"},{"key":"ref8","volume":"4503","author":"M Donier","year":"2014","journal-title":"A million metaorder analysis of market impact on the bitcoin"},{"key":"ref9","doi-asserted-by":"crossref","DOI":"10.1016\/j.eswa.2025.128867","article-title":"Optimizing market-making strategies: A multi-objective reinforcement learning approach with Pareto fronts","author":"\u00d3 Fern\u00e1ndez Vicente","year":"2026","journal-title":"Expert Systems with Applications 295, 128867"},{"key":"ref10","doi-asserted-by":"crossref","first-page":"2223","DOI":"10.1093\/rfs\/hhaa009","article-title":"Empirical asset pricing via machine learning","volume":"33","author":"S Gu","year":"2020","journal-title":"The Review of Financial Studies"},{"key":"ref11","first-page":"1321","article-title":"On calibration of modern neural networks","author":"C Guo","year":"2017","journal-title":"Proceedings of the 34th International Conference on Machine Learning"},{"key":"ref12","article-title":"A cautious multi-advisor sequential decision-making strategy without ground truth for maximizing the profits","volume":"286","author":"Z Guo","year":"2025","journal-title":"Expert Systems with Applications"},{"key":"ref13","first-page":"1861","article-title":"Soft actor-critic: Off-policy maximum entropy deep reinforcement learning with a stochastic actor","author":"T Haarnoja","year":"2018","journal-title":"Proceedings of the 35th International Conference on Machine Learning"},{"key":"ref14","doi-asserted-by":"crossref","first-page":"357","DOI":"10.2307\/1912559","article-title":"A new approach to the economic analysis of nonstationary time series and the business cycle","volume":"57","author":"J D Hamilton","year":"1989","journal-title":"Econometrica"},{"key":"ref15","article-title":"Feature enrichment imitative reinforcement learning for high-frequency trading","author":"F Han","year":"2026","journal-title":"Expert Systems with Applications 296, 129043"},{"key":"ref16","article-title":"Deep reinforcement learning that matters","author":"P Henderson","year":"2018","journal-title":"Proceedings of the Thirty-Second AAAI Conference on Artificial Intelligence"},{"key":"ref17","author":"E J Hu","year":"2021","journal-title":"LoRA: Low-rank adaptation of large language models"},{"key":"ref18","doi-asserted-by":"crossref","DOI":"10.1016\/j.eswa.2026.131776","article-title":"Beyond random walks: Exploring the learnability threshold of AI agents in algorithmic markets","volume":"315","author":"S K\u00fc\u00e7\u00fckoglu","year":"2026","journal-title":"Expert Systems with Applications"},{"key":"ref19","first-page":"2956","article-title":"Deep asymmetric multi-task feature learning","author":"H Lee","year":"2018","journal-title":"Proceedings of the 35th International Conference on Machine Learning, PMLR"},{"key":"ref20","author":"X Li","year":"2023","journal-title":"TradingGPT: Multi-agent system with layered memory and distinct characters for enhanced financial trading performance"},{"key":"ref21","doi-asserted-by":"crossref","first-page":"1748","DOI":"10.1016\/j.ijforecast.2021.03.012","article-title":"Temporal fusion transformers for interpretable multi-horizon time series forecasting","volume":"37","author":"B Lim","year":"2021","journal-title":"International Journal of Forecasting"},{"key":"ref22","author":"X Y Liu","year":"2020","journal-title":"FinRL: A deep reinforcement learning library for automated stock trading in quantitative finance"},{"key":"ref23","article-title":"iTransformer: Inverted transformers are effective for time series forecasting","author":"Y Liu","year":"2024","journal-title":"International Conference on Learning Representations"},{"key":"ref24","article-title":"A time series is worth 64 words: Long-term forecasting with transformers","author":"Y Nie","year":"2023","journal-title":"International Conference on Learning Representations"},{"key":"ref25","doi-asserted-by":"crossref","first-page":"4","DOI":"10.3905\/jpm.1988.409150","article-title":"The implementation shortfall: Paper versus reality","volume":"14","author":"A F Perold","year":"1988","journal-title":"The Journal of Portfolio Management"},{"key":"ref26","volume":"1","author":"N A Saidd","year":"2026","journal-title":"Decomposable reward modeling and realistic environment design for reinforcement learning-based Forex trading"},{"key":"ref27","article-title":"Hybrid reward architecture for reinforcement learning","author":"H Van Seijen","year":"2017","journal-title":"Advances in Neural Information Processing Systems"},{"key":"ref28","article-title":"Multi-task learning as multi-objective optimization","author":"O Sener","year":"2018","journal-title":"Advances in Neural Information Processing Systems"},{"key":"ref29","year":"2017","journal-title":"Customer advisory: Understand the risks of virtual currency trading"},{"key":"ref30","author":"G Woo","year":"2022","journal-title":"ETSformer: Exponential smoothing transformers for time-series forecasting"},{"key":"ref31","article-title":"Autoformer: Decomposition transformers with auto-correlation for long-term series forecasting","author":"H Wu","year":"2021","journal-title":"Advances in Neural Information Processing Systems"},{"key":"ref32","author":"Y Xiao","year":"2024","journal-title":"TradingAgents: Multi-agents LLM financial trading framework"},{"key":"ref33","author":"H Yang","year":"2023","journal-title":"FinGPT: Open-source financial large language models"},{"key":"ref34","article-title":"ReAct: Synergizing reasoning and acting in language models","author":"S Yao","year":"2023","journal-title":"International Conference on Learning Representations"},{"key":"ref35","article-title":"Gradient surgery for multi-task learning","author":"T Yu","year":"2020","journal-title":"Advances in Neural Information Processing Systems"},{"key":"ref36","volume":"13743","author":"Y Yu","year":"2023","journal-title":"FinMem: A performance-enhanced LLM trading agent with layered memory and character design"},{"key":"ref37","doi-asserted-by":"crossref","first-page":"3001","DOI":"10.1109\/TSP.2019.2907260","article-title":"DeepLOB: Deep convolutional neural networks for limit order books","volume":"67","author":"Z Zhang","year":"2019","journal-title":"IEEE Transactions on Signal Processing"}],"container-title":[],"original-title":[],"deposited":{"date-parts":[[2026,10,27]],"date-time":"2026-10-27T14:07:20Z","timestamp":1793110040000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.ssrn.com\/abstract=7674869"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026]]},"references-count":37,"URL":"https:\/\/doi.org\/10.2139\/ssrn.7674869","relation":{},"subject":[],"published":{"date-parts":[[2026]]},"subtype":"preprint"}}