NEWS & RESEARCH

Latest Updates

Insights, announcements, and research from the Vortex Legacy team.

Introducing Vortex Legacy: Research-Driven Technology for the Future

We are proud to announce the launch of Vortex Legacy — a technology company built at the intersection of rigorous academic research and production-grade engineering.

January 15, 2025

Research4 min read

Reading a Single-Factor Universe in the Correlation Engine

When a 12-coin watchlist produces an effective-number-of-bets of 2.4, what the panel is telling you and what the standard "diversify" intuition gets wrong about the situation.

May 4, 2026

Engineering4 min read

Why We Hand-Rolled Jacobi Instead of Pulling mathjs

The Correlation Engine ships its own ~150-line cyclic Jacobi eigendecomposition rather than depending on mathjs or numjs. The bundle-size and dependency-posture rationale, plus the numerical-stability tweaks the textbook algorithm omits.

May 4, 2026

Product Updates5 min read

Introducing the Correlation Engine: Six Regimes, Not One Matrix

The Correlation Engine partitions the observation window into six named regimes and recomputes the matrix on each. The choice of named regimes vs a continuous regime variable is deliberate — and the eigenvalue panel is what most readers actually want from a correlation tool.

May 4, 2026

Research5 min read

Tracing a Hypothetical Shock Through the Cascade Tensor

A descriptive walkthrough of the /simulate endpoint: a 5% downside shock seeded on a high-CSI asset, how the Jacobian operator propagates it wave-by-wave, and what "amplifier" vs "dampener" means in the resulting decomposition.

May 4, 2026

Engineering4 min read

Why CSI Blends Three Components (and Why Each Alone Is Insufficient)

The Cascade Stress Index combines density-at-risk, concentration, and skew into one scalar per asset. A walkthrough of why each component on its own misses the structural setup the others catch.

May 4, 2026

Product Updates5 min read

Introducing Cascade Tensor: Mapping Live Leverage Topology

Cascade Tensor builds a 4D position-density tensor across leveraged crypto venues, condenses it into the Cascade Stress Index per asset, and exposes the Jacobian of a linearised shock-propagation operator.

May 4, 2026

Research4 min read

Reading a Stable vs Near-Critical Capital Tensor Fit

α(A) is the headline stability number on a Capital Tensor snapshot. What an α(A) of −0.05 vs −0.001 vs +0.01 actually means, and why "stable" is always relative to the fit window.

May 4, 2026

Engineering3 min read

Why Capital Tensor Refuses Overlapping Category Inputs

A pre-fit guard catches the case where two CoinGecko categories share too many of their top-N coins. The choice of Jaccard 0.6 as the threshold is empirical — and fixes the documented 502-after-full-ingest failure mode.

May 4, 2026

Product Updates5 min read

Introducing Capital Tensor: From Money-as-a-Tensor to a Live Module

Capital Tensor implements the Money-as-a-Tensor framework against live capital data. Six panels — equilibrium, coupling, eigenvalues, dynamics, friction, Lyapunov — and what each one is asking.

May 4, 2026

Research5 min read

Reading a Diagnostics Score Across Regimes

A 70 in a risk-on regime and a 70 in a risk-off regime are not the same diagnosis. How regime detection re-weights the underlying composition, and why the score is meaningless without the regime context.

May 4, 2026

Engineering4 min read

Why Diagnostics Ships Two Normalisers (and Why You Should Never Compare Across Them)

Diagnostics uses MinMax normalisation on the single-coin /score path and Percentile normalisation on the universe /rank path. The two methods produce different numbers on the same data, and confusing them is the most common misread of the module.

May 4, 2026

Product Updates4 min read

Introducing Diagnostics: Why Ten Factor Groups

The reasoning behind the ten-factor structure of Diagnostics — why ten and not five or twenty, and how the three composite scores compose without double-counting.

May 4, 2026

Research3 min read

Poker-Inspired Stock Market Analysis in Python

A research notebook applying poker-style probabilistic reasoning — decision trees, binary-outcome frequencies — to historical stock-return sequences. For academic comparison, not trading advice.

March 1, 2024

Research5 min read

Sideways Markets: A Study of Indicators and Analytical Approaches

An analytical survey of technical indicators used to characterise stock markets that move laterally — descriptive research, not trading advice.

January 26, 2024

Research3 min read

Unifying Physics & AI for Stock Market Analysis

Exploring how ideas and theories from physics can be cross-fertilized with artificial intelligence to provide fresh perspectives on stock market dynamics.

August 23, 2023