About Vigil Analytics

Macro: Geopolitical Regime Classification

Vigil Analytics uses a quantitative model to detect the aggregated market response to global geopolitical conflicts. Instead of relying on qualitative news headlines, the model analyses the actual behaviour of oil futures, volatility indices, and commodity markets. These instruments move before and during conflicts in patterns that can be mathematically classified.

A Hidden Markov Model (HMM) processes three composite market signals daily: the oil calendar spread (near-term vs future prices), the OVX/VIX ratio (oil volatility relative to equity volatility), and the GVZ/VIX ratio (gold volatility relative to equity volatility). The model classifies the market into one of four geopolitical states:

• S1 — Active Escalation: A major geopolitical shock is unfolding. Oil in steep backwardation.
• S2 — Stalemate: Prolonged tension. Markets adapted to elevated baseline risk.
• S3 — De-escalation: Tensions easing. Risk premiums unwinding.
• S4 — Resolution: Conflict resolved. Normal market dynamics resume.

The model is causal — it only uses information available up to each day. Trained on 19 historical episodes from the Iraq War (2003) through the Russia-Ukraine war (2022) and the Iran-Israel escalation (2025), using the Baum-Welch algorithm with 25 random initialisations and BIC model selection.

CrisisFlow: Market Microstructure

CrisisFlow is a proprietary second-layer analytics tool that classifies market microstructure into three regimes: Compression (dealers net long gamma, moves dampened), Transition (inflection point, positioning shifting), and Amplification (dealers short gamma, forced hedging creates feedback loops).

Unlike the geopolitical HMM which uses a fixed transition matrix, CrisisFlow employs an Input-Output HMM (IO-HMM) where transition probabilities are conditioned on current signal values. This means the probability of transitioning from Compression to Amplification is higher when VVIX/VIX is elevated and the term slope is negative — a relationship the standard HMM cannot express.

The Regime Amplification Index (RAI) combines the IO-HMM posteriors with real dealer gamma exposure (GEX) and VIX to quantify structural cascade risk. High RAI means the model is confident the market is in a stress regime AND dealers are short gamma — a combination that creates self-reinforcing feedback loops.

The most powerful insight comes from combining both layers: the geopolitical overlay detects divergence — when the geopolitical model says calm but the microstructure says stress, or vice versa. These divergences often signal non-obvious risks.

Conflict Risk Scoring: Individual Stocks

Each S&P 500 stock receives a composite Conflict Risk Score (0 to 1) measuring its exposure to the current geopolitical crisis. The score combines four dimensions:

• Oil Price Sensitivity (30% weight) — Pearson correlation between the stock and WTI crude oil during S1 crisis periods. Energy stocks score high; software companies score low.
• Supply Chain & Shipping (30% weight)— A sector-level baseline reflecting structural supply chain exposure to the conflict region (e.g. Strait of Hormuz transit dependency, petrochemical feedstock reliance), adjusted by AI analysis of each company's specific geographic revenue and manufacturing footprint, then scaled by the current regime multiplier.
• Crisis Price Volatility (25% weight) — How much more volatile the stock becomes during S1 periods compared to normal. A stock moving 4x its normal range during crises scores 1.0.
• Crisis Volume Surge (15% weight) — How much trading volume increases during S1 periods. A stock trading 3x its normal volume during crises scores 1.0.

The supply chain dimension uses a hardcoded baseline covering 100+ GICS sectors and sub-industries across three conflict regions (Middle East, East Asia, Eastern Europe), then applies a DeepSeek AI adjustment of up to ±0.20 based on company-specific knowledge. An AI-generated explanation accompanies each score.

Coming Soon

Important Disclaimer

Vigil Analytics provides analytical output from quantitative models trained on historical market data. It does not constitute financial advice, investment recommendations, or an offer or solicitation to buy or sell any financial instrument. Past regime behaviour does not guarantee future performance. All investment decisions should be made with appropriate professional guidance and in the context of individual circumstances.