BiasMix-Finance: Post-Generation KYC Guardrails for LLM Portfolio Advice

A new arXiv paper (2608.28646v1) introduces BiasMix-Finance (Mini), a benchmark for testing post-generation KYC guardrails on LLM portfolio advice. Across three models and three inference modes, first-pass generations violated at least one cap in 47.6-85.7% of test cases (67.2% pooled), but a convex projection layer reduced final feasibility violations to 0% with a median correction distance of 0.066. The authors released the dataset, prompts, caps, and code in a public GitHub repository.

arXiv:2608.28646v1 Announce Type: new Abstract: Large language models LLMs can generate plausible-sounding ETF portfolios while silently violating basic KYC-style constraints on risk, fees, and diversification. This is especially problematic in agentic multi-turn advisory systems, where each draft recommendation can become an action unless guarded by an auditable enforcement layer. We study a model-agnostic, asset-agnostic post-generation guardrail pipeline: i enforce a strict JSON allocation schema, ii validate allocations against numeric caps, and iii when violations occur, deterministically project the output to the nearest feasible portfolio via a convex quadratic program QCQP . We introduce BiasMix-Finance Mini , a compact stress-test benchmark for constrained decision-making under biased LLM generations, with a 16-ETF universe, three investor profiles, and eight bias prompts. Across three models and three inference modes direct, critique, self-consistency , first-pass generations violate at least one cap in 47.6-85.7% of test cases 67.2% pooled , but the convex projection layer reduces final feasibility violations to 0% while requiring only a small correction distance test pooled median D=||w -w0|| 2=0.066 , indicating that the guardrail typically preserves the intent of the original allocation. We report violation rates and correction distances with confidence intervals, and paired model comparisons with multiple-testing correction. To support reproducibility, we release the dataset, prompts, caps, and code in our public GitHub repository.