Research

Reflexivity Kernel Spectroscopy

SSRN Abstract ID 6450561

Separates how much of a market's movement is self-reinforcing feedback versus genuine outside news, reaching 76-83% out-of-sample accuracy.

Abstract. Introduces RKS as a constrained estimator of the reflexivity kernel, decomposing realized covariance into endogenous flow-driven risk and exogenous innovation residuals. The paper reports admissibility-certified calibration evidence, 76-83% out-of-sample R-squared, and 75-82% reflexivity share while surfacing failure artifacts instead of silent fallback.

Constraint Shadow-Price Tomography

SSRN Abstract ID 6457180

Reconstructs hidden funding and balance-sheet stress across markets from small pricing gaps in FX, Treasuries, and swaps. The signal peaked in the March 2020 crisis.

Abstract. CSPT inverts persistent CIP, Treasury basis, and swap-spread wedges into a time-versioned Market Dual State that measures the implied shadow prices of binding balance-sheet, funding, and margin constraints. The March 2020 study recovers a peak balance-sheet coordinate of 152.91, quarter-end identification at p = 0.000777, and held-out swap-spread pricing error as low as 3.10 bp.

Epistemic Curvature

SSRN Abstract ID 6523041

Under review, Computational Economics (Springer)

Pinpoints which market expectations are most fragile before volatility hits, an early-warning signal that leads the VIX. Under review at a Springer journal.

Abstract. EC measures the Riemannian curvature of the Fisher-Rao manifold of market-implied probability distributions, extracting an Epistemic Curvature Spectrum of fragile and robust belief directions. The April 2026 paper reports 62/62 unit tests, 5/5 falsification tests, option-implied manifold construction, and predictive evidence that curvature leads VIX and return-implied curvature across regimes.

Admissible World Deformation Field

SSRN Abstract ID 6524938

Measures how internally contradictory current market prices are, and pinpoints exactly which assumptions are breaking.

Abstract. AWDF solves a constrained quadratic program to compute the minimum deformation from observed cross-asset states into the nearest admissible world, separating hard structural contradictions from softer plausibility pressure through KKT dual tension. The April 2026 paper reports a full falsification pass rate, reproducibility of 1.0, stability of 0.963 under perturbation, and live-public identification of the binding fracture stack behind structural inconsistency.

Irreversibility Field Anatomy

SSRN Abstract ID 6606258

Borrows physics, specifically thermodynamics, to tell friction-driven crises apart from information-driven ones, and sets a hard ceiling on how much a strategy can earn.

Abstract. IFA models equity markets as non-equilibrium systems, constructing probability current fields, entropy production rates, and a Hatano-Sasa housekeeping/excess decomposition to classify crisis mechanics. The paper reports 104/104 pre-registered falsification tests, separates friction-dominated from information-dominated crises, and derives a Thermodynamic Uncertainty Relation ceiling for linear strategy Sharpe capacity.

Gyral Covariance Decomposition

SSRN Abstract ID 6597020

Under review, Computational Economics (Springer)

Splits how stocks move together into a stable part and a rotating 'circulatory' part, and finds the circulatory part is priced across the Russell 1000. Under review at a Springer journal.

Abstract. GCD decomposes the drift matrix of a multivariate Ornstein-Uhlenbeck process into symmetric gradient drift and antisymmetric circulatory drift, yielding a closed-form circulatory covariance perturbation. Across five equity universes, the paper finds persistent 45-55% circulatory covariance shares, decisive permutation evidence, VIX stress-peak detection, and a significant Russell 1000 anti-premium on circulatory beta.

Executable World Resolution

SSRN Abstract ID 6639600

A decision framework that tells you whether a trading idea is actually testable with the assets you can trade today, turning research into clear accept, reroute, or kill calls.

Abstract. EWR defines a capital separation function, resolution tensor, and executable partition that measure the deployable cost of distinguishing future worlds under liquidity, margin, financing, and concentration constraints. The paper converts pre-forecast research into ACCEPT, REROUTE, KILL, and UNIDENTIFIED governance decisions, defeating classical span diagnostics in matched stress-regime studies.

Capital Trigger Topology

SSRN Abstract ID 6715359

Predicts where big pools of capital are mechanically forced to trade, using only public rules and filings. Called direction correctly in 8 of 8 test cases.

Abstract. CTT reconstructs the rule that compels action rather than estimating it statistically, instantiated first on the leveraged and inverse ETF daily-reset automaton across an append-only registry of 18 certified ProShares products with SHA-256 provenance. Replaying the August 2015 and February 2018 information cutoffs against SEC EDGAR filings, realized signed VIX-futures position deltas match the predicted episode-level direction in 8 of 8 ticker-episode pairs at quarter-end anchors. The paper reports its failure honestly: the pre-registered daily-resolution flagship gate is uncrossed and a hard-kill artifact is emitted, with 43 tests reproducing both replays byte for byte.

Vintaged Capital Ledger

SSRN Abstract ID 6781319

Reconstructs who still holds an asset and at what cost basis, tested exactly against 189M+ Bitcoin records, then bounded under real disclosure limits.

Abstract. VCL treats the installed-capital population as the primary state object, observable exactly in Bitcoin UTXO data and partially identified under disclosure-bound observation. Building a short-term-holder oracle over 189,315,387 transaction outputs and imposing a monthly grid that mimics SEC NPORT-P cadence, the August 2015 cutoff yields an identified-set width of 0.375 and February 2018 yields 0.672. A pre-registered sweep across 15 to 90 day cadences shows monthly fidelity is insufficient in the high-volatility 2017 regime while quarterly-or-coarser cadence clears the conditional gate, with mass conservation at machine epsilon and 133 unit tests passing.

Scheduled Coordination Density

SSRN Abstract ID 6818821

Measures when large scheduled trades pile into the same closing window, and shows it predicts overnight returns across 15 ETFs over 8 years (p = 0.000).

Abstract. SCD is built only from public calendars and public price and volume data, decomposing exactly into a self-mass term and a cross-clock overlap term whose normalized form is the object of study. Across a fifteen-ETF panel from 2016 to 2024 (32,130 ETF-days) the overlap coefficient sits at the 0th percentile of a 1,000-iteration calendar-randomization null (p = 0.000). Pre-registered against the opposite hypothesis, higher coordination predicts overnight return continuation rather than reversal (-5.03 bp per standard deviation, t = -2.61), an effect specific to the modern leveraged-ETF and short-dated-option era, with four directional hypotheses rejected and committed as first-class artifacts.

The Mandated Revelation Field

SSRN Abstract ID 6868499

Tracks facts that are already legally true but not yet public: a dated map of information the market is required to reveal, built from 1,132 SEC filings.

Abstract. MRF splits into an exact timing layer computed deterministically from public regulatory rules and a magnitude layer reported as Manski partial-identification bounds rather than point estimates. The reference implementation runs on free public data: a hashed lattice of eleven rule branches across four statute families, an EDGAR ingest of 1,132 filings, and a dated owed-fact ledger of 2,583 facts with a byte-identical canonical hash, all enforced by four integrity gates in a 94-test suite. Exploiting the February 2024 Schedule 13D deadline change as a natural experiment, the pre-registered flagship return-prediction claim fails out of sample and is reported dead verbatim, while the orthogonalized cross-statute component remains incrementally significant (t = 3.09, p = 0.002).

Climate Alpha Working Paper

A practical blueprint for turning physical climate risk into portfolio signals, position limits, and risk controls.

Abstract. Frames physical climate risk as an upstream state variable for portfolio construction and strategy workflows, translating scenario evidence into decision-grade mitigation and allocation outputs.