This work presents a chunked zero-knowledge circuit design that attests an aggregated fairness statistic - the demographic-parity gap - on committed, nonce-sampled batches of real 2022 HMDA mortgage data, and presents a fuller mapping from SR 11-7 and OCC 2011-12 control language to zero-knowledge statements.
Abstract
Deploying ML models in regulated decision-making (credit underwriting, fraud detection, loan approval) requires demonstrating fairness and robustness to auditors without exposing model weights or customer data. We address this attestation problem for U.S. bank supervision under SR 11-7 and OCC 2011-12 guidance. We present a chunked zero-knowledge circuit design that attests an aggregated fairness statistic - the demographic-parity gap - on committed, nonce-sampled batches of real 2022 HMDA mortgage data, and demonstrate it end-to-end: 32,768 rows, 32 independently verified zkSNARK proofs, aggregated attested gap within 0.0029 of the true held-out value, per-chunk proving under 4 seconds. We also demonstrate extensibility by attesting a second control on the identical architecture - expected calibration error at 10 bins - with all 32 chunks verified, per-chunk proving at about 14.7 seconds, and attested ECE within 0.00037 of the plaintext value on the same committed rows. We compare this design against two alternatives - a flat summation circuit, which overflows past a few thousand rows, and a tree-reduction circuit, numerically exact but intractable to compile - and find the chunked design is the only one that reached this scale. We discovered and root-caused a genuine data-quality failure (a sentinel-code outlier distorting circuit proving and the fairness statistic itself) and resolved it with a published preprocessing specification. We also propose a fuller mapping from SR 11-7 and OCC 2011-12 control language to zero-knowledge statements (nine control elements spanning soundness, calibration, robustness, and drift), a nonce-based sampling protocol resisting bank-side cherry-picking, and a threat model for deployment, proposed as design work, not implemented results. Two controls, one model class, one task are demonstrated end-to-end; the rest is scoped and left as future work.
Regulation (EU) 2024/886 obliges European payment service providers to settle euro credit transfers in under ten seconds, around the clock. This removes both the overnight batch window in which anti-money-laundering (AML) analytics traditionally ran and the settlement delay that made recovery possible, forcing detection, explanation and decision inside a single-digit-second envelope. We present KONTOGRAPH, an end-to-end AML pipeline for the SEPA Instant rail built under a self-imposed 200 ms 99th-percentile budget, and report an empirical study on 1,562,860 simulated payments with injected typologies and deliberately incomplete labels. Three findings are of interest beyond the system itself. First, a temporal graph network with per-node memory improves PR-AUC over a gradient-boosted tabular baseline from 0.0053 to 0.1717, a paired day-blocked bootstrap difference of +0.166 with 95% CI [0.105, 0.241]; per-node memory alone more than doubles the score. Second, expressing each feature once and compiling it to three execution backends, with equivalence enforced by property-based tests that perturb the future, surfaced three point-in-time violations that code review had passed--each of which would have inflated reported performance. Third, and most consequential for practice, exporting the deployed tree ensemble to ONNX changed only $7.4 \times 10^{-8}$ in mean score yet altered 0.26% of decisions and inflated the alert volume by 12%, because 32-bit accumulation perturbs scores across a cost-optimal threshold of $3.98 \times 10^{-4}$. We argue that a serving-format conversion must be treated as a model change until measured, and that fidelity metrics for subgraph explainers can be vacuous when candidate neighbourhoods are small--a null result we report in full.
None of the step-level credit signals the authors audit -- LLM-judge scores, outcome-conditioned logprob ratios, or the policy's own confidence -- shows reliable incremental fidelity beyond its own marginal-matched shuffled control.
Corporate annual reports contain weakly structured evidence about foreign-exchange risk management, derivative use, natural hedging, and explicit non-use. This study develops AWARE-FX, an auditable AI/NLP decision-support system that converts report text into traceable firm-year hedging-disclosure measures. The system combines a professional-source lexicon, negation and accounting-status logic, channel-specific financial encoders, exact evidence gates, conservative aggregation, and an audit ledger. Across 24,909 Hong Kong firm-years from 2008-2025, it retrieves and scores 543,527 snippets. Reliability is evaluated through ablations, a stratified 300-snippet human audit, three-seed FinBERT-ModernBERT comparisons, strict 2023-2025 temporal tests, probability calibration, selective prediction, and fixed-prompt generative-model benchmarks. FinBERT has the higher mean F1 in seven of eight encoder task-split comparisons; its temporal F1 ranges from 0.702 to 0.872. Abstaining on the 20% least-confident temporal observations raises retained-sample F1 by 0.050-0.077. Deterministic Qwen3-8B performs strongly on commodity and negation evidence but poorly on foreign-debt and accounting-context labels, showing that a general-purpose LLM does not uniformly replace domain constraints. The strict FX score is negatively associated with linked baseline and stress-period FX exposure, whereas the generic broad score is not. These associations provide external construct validation, not causal estimates of hedging effectiveness. AWARE-FX contributes a tested decision-support architecture in which retrieval, status logic, classification, uncertainty handling, aggregation, and external validation remain separately auditable.
OpenMarket began as an attempt to trade Polymarket's BTC 15-minute binary markets against Binance BTC/USDT order flow. The attempt did not produce a tradable edge: out-of-sample, a walk-forward logistic model over 43 microstructure features does not beat, and slightly underperforms, the probability already implied by Polymarket's own order book, and simulated trading nets -0.116 normalized payoff units per attempted trade under stated fee and slippage assumptions. We release the synchronized corpus and infrastructure that attempt produced and, to our knowledge, the first public millisecond-level Polymarket BTC / Binance BTC-USDT paired corpus with explicit pairing metadata. The frozen archive (tag v0.5.2) contains 727,098,247 deduplicated rows across 202 archival snapshots, with event data on 54 observed Polymarket days (57 Binance days) between 2026-02-12 and 2026-05-15, including 2,936,031 explicit lead-lag pairs, alongside a reproducible Rust pipeline for collection, millisecond pairing, Parquet export, and walk-forward calibration. Initial analyses establish Polymarket stylized facts (one-tick top-of-book spreads) and characterize cross-venue timing: an apparent 16 ms median lag on venue source clocks, with relative clock drift bounded to at most 6 ms over the archive but a remaining single-vantage constant-offset ambiguity of approximately plus or minus 99 ms. A synchronization-free event study, measured only on the collector clock, independently shows that Polymarket quotes respond to large Binance moves after a median 347 ms. We position this work as a data-and-methods release whose central empirical result is a null out-of-sample forecasting result.
Every production model is updated, by retraining, fine-tuning, quantization, or a silent vendor swap, and each update risks being worse than what it replaced. We formalize update promotion as certified paired risk-difference auditing. Our starting point is a support identity: the risk difference between two models lives on the inputs where they disagree, observable without labels. We build DISCERN, a sequential two-tier protocol. A zero-label tier certifies benign updates whose disagreement rate is below tolerance from unlabeled traffic alone. An audited tier labels only sampled disagreements through an anytime-valid confidence sequence, valid at every stopping time and under any label-routing rule, even an adversarial judge. We prove finite-sample validity and matching label-complexity bounds of order rho^2/eps^2 at the rate level, so exploiting free disagreement provably saves a factor 1/rho over any pairing-blind auditor, and the guarantee composes across an unbounded sequence of promotions from one error budget. Across 14,000+ replayed audit streams over 785 update pairs, including LoRA fine-tunes of language models up to 1.4B parameters, miscoverage is 0.0002 (nominal 5%), power 0.986 with zero false alarms, and 56% of benign updates certify with zero labels. Each audit emits a machine-checkable evidence record for post-market monitoring.
Costly LLM features matter only if calibration lets them affect the forecast. We document a failure of this link in a next-day risk study of two broad-market funds. Full-history scoring preceded the 2022 calibration. Calibration then set all four LLM weights to zero. The 856 later scores therefore could not affect the evaluation. We call this calibration-induced degeneracy. Allowing signed weights reactivated all four mappings. None improved forecasts after familywise correction. By contrast, a near-zero-cost headline count reduced SPY variance-forecast loss by 0.001720 (95 percent familywise interval: [0.000719, 0.002830]). The cheap baseline is therefore a critical diagnostic. We propose a calibration-viability checkpoint. Fit the mapping, perturb the feature over prespecified calibration values, and require a meaningful forecast response before acquiring holdout features. The check uses no holdout outcomes. Here, it would have stopped the paid full-history inference phase.
Arin Mohanty· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
