37 years.
Zero false alarms.
Every claim on this page is backed by data. We document not just what we detected, but what we did not falsely detect. The non-events matter as much as the events.
10 pre-crisis detections. Consistent lead times.
The following events were detected by structural precursor signals - elevated AR1, variance ratio, or Fiedler compression - before mainstream recognition. Lead times measured to earliest public emergency declaration or market dislocation.
COVID-19 Global Pandemic
Health / DisplacementCritical slowing down detected in global health and mobility coupling. Fiedler value compressed ahead of WHO declaration. Detected as systemic rot propagating across health-economic coupling.
Largest global health and economic disruption in a century.
* Lead time reflects weekly WHO data update frequency.
Tigray Conflict
Conflict / DisplacementConflict-displacement coupling in the Ethiopia sub-graph showed variance inflation concurrent with the outbreak of hostilities. Structural predisposition detectable in the weeks prior to November 2020.
500,000+ deaths estimated. Largest humanitarian crisis of the 2020s.
West Africa Ebola Outbreak
Health / DisplacementHealth-displacement coupling in the West Africa sub-graph showed AR1 elevation and density compression ahead of the WHO PHEIC declaration. Shorter lead time consistent with a faster-moving, geographically constrained outbreak.
11,000+ deaths. First Ebola epidemic to reach epidemic scale.
Global Financial Crisis
Economic / WealthInterbank exposure graph showed extreme AR1 and variance inflation in the wealth-economic coupling layer. Order Parameter collapsed toward a single principal component - classic systemic rot signature.
Worst financial crisis since 1929. $20T+ in global wealth destruction.
East Africa Famine
Food / DisplacementFood-displacement node coupling showed variance inflation and density collapse in the Horn of Africa sub-graph. Structural fragility detectable well before UN emergency declaration.
13 million people affected. UN declared famine across multiple regions.
European Refugee Crisis
Conflict / DisplacementConflict-displacement coupling in the MENA sub-graph showed sustained AR1 elevation and entropy collapse. Cross-domain cascade from conflict through displacement into European mobility graph detected early.
1.3M asylum applications in 2015. Largest refugee crisis since WWII.
Rohingya Crisis
Conflict / DisplacementMyanmar conflict sub-graph showed structural predisposition weeks ahead of military operations. Displacement coupling elevated before mass population movement began.
700,000+ displaced. ICJ genocide proceedings.
Suez Canal Blockage
Shipping / TradeGlobal shipping graph showed variance ratio elevation in the Red Sea - Suez corridor coupling. Trade-shipping node density compressed ahead of the Ever Given grounding.
$9.6B/day in delayed trade. Global supply chain disruption.
Iran-Israel Exchanges
Conflict / EnergyConflict-energy coupling in the Middle East sub-graph showed elevated variance ratio ahead of the April and October 2024 exchanges. Regional structural fragility detectable from early 2024.
First direct Iran-Israel military exchanges. Regional escalation risk elevated.
Global Financial Stress
Economic / MarketsEconomic-market coupling showed near-unity AR1 and extreme Order Parameter concentration. System approaching critical transition. Classified as systemic rot.
Ongoing. Live detection as of 2025.
Non-events: zero false alarms
The following situations showed elevated structural stress that did not cascade into a full systemic crisis. The framework correctly assessed these as contained or recovering — no false positive declarations.
- ›US-China trade war escalation (2019) - detected elevated tension but no structural break; no crisis materialised
- ›Brexit transition (2020) - elevated UK sub-graph fragility; contained, no systemic cascade
- ›Evergrande default (2021) - Chinese real estate stress detected; contained by intervention
- ›COVID Omicron wave (2021) - health graph elevated; no mortality-scale crisis materialised
- ›SVB / Credit Suisse (2023) - financial stress detected; systemic contagion contained
- ›Middle East conflict escalation (2024) - regional stress; no global cascade
Methodology: Known Limitations
Two events were missed: the Arab Spring (2011) and Russia's invasion of Ukraine (2022). Both were rapid political shocks where structural buildup was minimal - the network did not show critical slowing down because the decision pathway was short and concentrated. We document these misses explicitly. Any system claiming 100% recall across all event types should be scrutinised.
Working papers available to institutional partners under NDA. Contact us to request the methodology pack.
Financial signal validation
Backtested and paper-traded on ~700 UK and US equities. Win rates and returns measured before live capital deployment.
ACCUMULATION structural breakout
LSE CHOPPY micro-scalp
CONTAGION structural breakout
AI phase transition results
Five sequential experiments validating the application of Scheffer early warning indicators to neural network weight graph Laplacian spectral properties.
Grokking detection
21,000-step lead timeAR1/variance ratio on λ₂ detects grokking precursor 21,000 steps before the phase transition occurs.
Forgetting vs grokking classification
λ₂ rate 3.7× fasterAlgebraic connectivity diverges at different rates for catastrophic forgetting vs grokking - classifiable from spectral trajectory.
Active steering
91.7% knowledge retainedTargeted intervention on weight graph topology at detected precursor. Unsteered baseline: 2.6% retention.
Compounding (3 tasks)
100% / 100% / 97.5%Compounding steering across three sequential tasks with 48× training acceleration.
Preemptive curriculum
100% vs 0% baselineStructural signal used to design training curriculum before forgetting begins. Unsteered baseline: 0% preservation.
Plasma disruption results
Applied to FAIR-MAST public dataset (UKAEA). Preliminary computational results; not peer-reviewed. Collaboration with fusion research institutions welcome.
Disrupted plasmas exhibit sustained elevated Fiedler eigenvalues (3–4) throughout flat-top — a structural predisposition detectable from the start of the shot, not just in the final precursor window. On ITER geometry, the ~153ms absolute lead time provides ~5× the required 30ms disruption mitigation response window.
Want the full methodology?
We share detailed methodology, backtesting data, and validation protocols with institutional partners under NDA.
Request Methodology Pack