A Universal, Options-Free, One-Byte Market-State Primitive: Cross-Asset Transfer, Distributional-Novelty Detection, and Privacy-Preserving Federation

Abstract

A Universal, Options-Free, One-Byte Market-State Primitive: Cross-Asset Transfer, Distributional-Novelty Detection, and Privacy-Preserving Federation Randolph James Ferlic, M.D. and Kimberly Kate Ferlic (Fieldstone Analytics, LLC, Austin, TX, USA) Preprint · Zenodo DOI: 10.5281/zenodo.22116173 · CC-BY 4.0 · Community: spiral-domain-encoder-campaign Abstract A companion study established that a fixed class-discriminant encoder — which reduces a window of a multivariate stream to a single 8-bit token — is a near-lossless detector of market stress across asset classes, but a taxed classifier and a null forecaster. This study asks, under strict pre-registration, whether that detector becomes a *universal, deployable primitive*, and finds that it does, along five axes, with an honest boundary preserved throughout. (1) The codebook is universal: fit on equities alone, it detects stress zero-shot across cryptocurrency, commodities, emerging-market funds, credit, and foreign exchange at a median AUC of 0.84, within 0.01 of a bespoke per-asset codebook, because the calm-to-stress discriminant direction is nearly identical across asset classes (median cross-asset cosine 0.94). (2) The token detects distributional regime-change that volatility structurally misses — the one place it beats volatility — and is an orthogonal channel that improves a volatility-only stack by +0.18 AUC on a mixed target. (3) It generalizes to credit and rates (pooled AUC 0.91) and to currency crosses (0.80), including markets with no options index. (4) It is cheap at scale and low-data: a genuine one-byte-on-the-wire firm-wide dashboard detects systemic stress at 0.92 at 200× compression — a 42-instrument universe fits in ~10 kilobytes per year — and a new deployment is usable after ~3 months of calm history. (5) The same one-byte token supports privacy-preserving federation: institutions each sharing one token per day match full-data systemic-detection utility while their individual positions are unrecoverable even to a non-linear, colluding, market-informed adversary (incremental reconstruction R² ≈ 0.01). All headline results survive a permutation-placebo leakage test, moving-block bootstrap confidence intervals, and robustness to split, seed, stress-label, and a strict recent holdout. Every negative is reported, and two honest scope refinements are disclosed: a formal differential-privacy guarantee is event-level, not stream-level, and portfolio membership is coarsely inferable while positions are not. Highlights · Universal codebook — one equity-fit codebook detects stress zero-shot across every asset class (median 0.84, transfer loss ~0.01, 22/23 targets); mechanism = a shared cross-asset stress direction (cosine 0.94). · A capability volatility lacks — the token detects distributional (shape) regime-change where trailing volatility is at or below chance; an orthogonal channel that complements any GARCH/VIX stack. · Coverage — credit/rates (pooled 0.91) and FX/carry white-space (0.80), options-free, at one byte per instrument. · Compression-at-scale — a 1-byte-on-the-wire firm-wide dashboard at AUC 0.92, 200× compression, ~10 KB/instrument-year; detection-sufficient encoder ≈ 590 multiply-accumulates and < 3 KB (microcontroller-class). · Low-data cold-start — usable after ~3 months of calm data; saturates within ~2 years; train on a trailing window. · Privacy-preserving federation — full-data-equivalent systemic detection with individual positions non-invertible even under a strong adversary; optional event-level differential privacy; manipulation-robust median aggregation; per-channel attribution; a yield-curve-shape token. · Information-theoretic characterization — detection is a ~1-bit decision while classification is high-bit; competitive with principled two-sample/OOD tests at O(1) online cost. · Honest negatives and hardening — network topology, temporal-token, cross-sectional ranking, and tail-edge results are all reported null; every headline result is leakage-tested (permutation placebo → chance) and confidence-bounded. What this record contains · Manuscript_Paper40.pdf — the manuscript (single-column preprint). · PAPER_40_ZENODO_ARCHIVE.zip — the reproducibility archive: seven pre-registration scopes (frozen outcome bands) covering studies FIN-15 … FIN-45; the deterministic per-study runners and the shared frozen encoder/loader and federation modules; the per-study JSON result summaries behind every figure and table value; the figure-rebuild script; and the eight figures. All paths and identifiers are scrubbed and leak-scanned per the campaign deposit discipline; no raw market data is redistributed. Cite as R. J. Ferlic and K. K. Ferlic, "A universal, options-free, one-byte market-state primitive: cross-asset transfer, distributional-novelty detection, and privacy-preserving federation," Zenodo, 2026, doi: 10.5281/zenodo.22116173. License and patent notice Released under the Creative Commons Attribution 4.0 International License (CC-BY 4.0). Consistent with that license, no patent, patent-application, or other intellectual-property right of the authors is licensed, waived, granted, or otherwise conveyed by this deposit; the methods described — including the single-token class-discriminant encoder, its multi-token product-quantization variant, its unsupervised nearest-centroid-distance monitoring mode, and the privacy-preserving multi-party aggregation of its tokens — are the subject of filed and pending U.S. patent applications. Licensing inquiries: randolphf@fieldstoneanalyticsllc.com. Companion deposits (spiral-domain-encoder-campaign) · Single-token financial market-state monitor (companion study): doi:10.5281/zenodo.22101085 · Single-token industrial sensor substrate: doi:10.5281/zenodo.20854722 · Hardening and generality characterization: doi:10.5281/zenodo.20802759 · Deterministic multi-token token ladder: doi:10.5281/zenodo.22003179 Keywords universal codebook; decision-preserving compression; distributional novelty; federated privacy; differential privacy; systemic risk; market-stress detection; class-discriminant codebook; cross-asset transfer; edge computing; anomaly detection; pre-registration; honest negatives; financial time series; regime detection; credit risk; yield curve; volatility regime; VIX; market surveillance