TITLE:
Filter before You Solve: A Deterministic-First/Learned-Second Architecture for AI-Driven Portfolio Management with Real-Money Training-Investment Calibration
AUTHORS:
George Melville, Dena Ghiassi, Scott Inthathirath, Julian Yeomans
KEYWORDS:
Explainable AI in Finance, AI-Driven Strategy, Machine Learning Insight AI in Business, Options Management, Filter-before-You-Solve, Deterministic-First/Learned-Second Architecture, Hallucination-Free, Interpretability, SimDec, Real-Money Training
JOURNAL NAME:
Journal of Software Engineering and Applications,
Vol.19 No.8,
August
26,
2026
ABSTRACT: This study investigates a deterministic-first/learned-second AI/ML framework that is deployed in regulated retail brokerage accounts. A two-stage calibration system is employed that combines historical back-testing with live recalibration via continuous position snapshots. The framework includes a unique explainability layer that employs the global sensitivity analysis method, SimDec, to identify the most influential components. SimDec renders every admissible decision fully attributable to a joint state of the input space. Because the attribution is an input to the trigger rather than a commentary on it, no generative step exists in the decision path and hallucination is structurally unreachable and, therefore, absent by construction. An intraday options trading financial framework is illustrated through a live training-investment cycle on long-call positions using “real money”. The major contribution of this research is the overall architectural and methodological framework. The filter-before-you-solve approach enables contributions to be evaluated independently of specific implementations. Beyond financial applications, it is described how the complete architectural pattern can actually generalize to many AI/ML deployment contexts that require auditable deterministic gating prior to learned inference.