Research Prototype — v6.0
April 2026

Know what your
portfolio is
actually exposed to.

SRX translates macro systemic stress into portfolio-specific risk intelligence — which holdings amplify exposure, how diversification is degrading, and when regime conditions are shifting.

0d
Early warning before Lehman — 2008 GFC

GSRI crossed Elevated on Sept 27, 2007 — 376 trading days before Lehman.
VIX crossed 30 on Nov 12, 2007. GSRI was 69 days earlier.

13
Periods Validated
11 crisis · 2 calm baselines
0
Calm False Positives
2017 peak GSRI: 8.7
+69d
GSRI vs VIX
Earlier on 2008 GFC
GSRI — Global Systemic Risk Index / 10-Year History
Live · Monitoring
The Gap

Standard risk models are blind to regime shifts.

VIX tells you the market is stressed. It doesn't tell you what that stress means for your specific holdings — which positions are amplifying exposure, whether your diversification is still working, or how far a regime shift has progressed.

Most portfolio risk tools are calibrated to normal conditions. During structural decay — the slow buildup of cross-asset correlation and drawdown acceleration that precedes major dislocations — they produce misleading signals.

SRX answers a different question: not how volatile is the market, but what is the current systemic regime doing to this portfolio, specifically.

Platform
ModuleWhat it measuresWhat it answers
GSRI Monitor
Equity vol (GARCH), credit stress, equity-credit co-movement, treasury vol, crypto vol, directional cross-asset correlation — aggregated into a 0–100 regime score with ML-optimized component weights.
Where is systemic stress right now, and is it building or receding?
SRS Engine
Four signals — volatility, directional correlation, drawdown, liquidity — with SLSQP-optimized weights (L2 regularization, 13-era LOOCV), convex non-linear scoring, and headroom-proportional modifiers.
Which stress channels are active and how severe is each?
Portfolio Impact
Core Differentiator
Weighted systemic contribution per holding, diversification fragility, vol sensitivity, drawdown exposure — translated from macro GSRI/SRS signals into holding-specific consequences.
What does the current regime mean for this specific portfolio?
Stress Testing
2008 GFC, COVID crash, rate shock, liquidity freeze, crypto contagion — scenario-mapped to portfolio positions with per-asset loss attribution and stressed PRI recomputation.
How would this portfolio behave under each historical stress regime?
Shock Detector
5-day acceleration in vol and drawdown — 1.25× multiplier on sudden dislocations. Regime classified as structural decay, exogenous shock, or normal. Adaptive window compresses under sustained stress.
Is this slow structural decay or a sudden exogenous shock?
ML Weight Optimizer
SLSQP constrained optimization with L2 regularization against 13 historical eras. Leave-one-era-out cross-validation. Kalman filter threshold calibration on 10-year live signal history.
What signal weights best detect crises while staying quiet during calm?
Historical Similarity
Cosine similarity between current 6-component signal vector and labeled historical regime snapshots (2008, 2020, 2022, SVB, EU Sovereign, and more).
Which historical period does the current environment most resemble?
Systemic Risk Monitor01
GSRI + SRS regime dashboard
Portfolio Impact Layer02
Per-holding stress attribution
Scenario Stress Testing03
Historical crisis mapping
Historical Validation04
Walk-forward — 13 periods tested
Signal Architecture

Six signals.
One regime score.

Each signal uses a convex power function (γ > 1) that accelerates under stress. The correlation signal applies directional conditioning — only downside co-movement scores high, preventing risk-on rallies from triggering false positives. Signal weights are optimized via SLSQP with L2 regularization against 13 historical crisis eras, with leave-one-era-out cross-validation. Scoring thresholds are calibrated via Kalman filter on 10-year live signal history.

SRS
= λV·ϕ(Vt) + λL·ϕ(Lt) + λ·ϕ(C̃t) + λM·ϕ(Mt)
λ optimized via SLSQP · L2 regularization · 13-era LOOCV · Kalman threshold calibration
Vtλ = 0.30*
Volatility

GARCH(1,1) conditional vol per asset; rolling std fallback. Calm: 5% ann. Crisis: 60% ann. GSRI uses per-asset GARCH; SRS uses cross-asset mean.

γ = 1.5
Ltλ = 0.05*
Liquidity

Amihud illiquidity ratio. Low weight — Yahoo Finance ETF volume conflates primary and secondary market activity. Bounds [0.02, 0.60] in ML optimizer.

γ = 1.3
C̃tλ = 0.35*
Correlation

Directional pairwise ρ — scaled by fraction of assets with negative mean returns. BTC excluded. SPY 120-day MA trend conditioned. Strongest structural signal.

γ = 1.4
Mtλ = 0.30*
Drawdown

Avg. drawdown from rolling peaks. Highest convexity exponent — crisis non-linearity is sharpest here. * Heuristic prior; ML optimizer finds data-driven weights.

γ = 1.6
Portfolio Translation
Key differentiator

The SRS and GSRI are macro signals. SRX's Portfolio Impact layer translates them into holding-specific consequences: which assets amplify systemic exposure relative to their weight, whether cross-asset diversification is structurally intact, and what the dominant stress transmission channel is for this portfolio. EC and VTS modifiers use headroom-proportional scaling so they never inflate already-elevated base scores. One-click PDF reports embed all readings alongside AI-generated regime narratives and historical similarity analysis for client-facing use.

Historical Validation

Walk-forward protocol.
Zero look-ahead bias.

Thresholds fixed a priori. Same parameters across all 13 periods. Not re-calibrated per event. Eleven crisis detections · Two calm baselines · Thirteen eras total.

2008Structural Decay
Global Financial Crisis
Sept 15, 2008 — Lehman collapse
Lead Time
376 trading days
vs VIX / MOVE (Calibrated)
GSRI +69d earlier than VIX
Primary Driver
Directional correlation C̃t + equity-credit co-movement. First Elevated crossing Sept 27, 2007 — after Northern Rock, BNP Paribas freeze, first Fed emergency cut.
Analyst Note
Primary validation result. Structural decay detected 17 months before the acute crisis. VIX crossed 30 on Nov 12, 2007. GSRI outperforms VIX and MOVE at calibrated equal-FPR thresholds.
2020Exogenous Shock
COVID-19 Crash
Mar 11, 2020 — WHO Declaration
Lead Time
Simultaneous with VIX
vs VIX / MOVE (Calibrated)
Same day at fixed thresholds
Primary Driver
Equity vol doubled within 5 days triggering the 1.25× Shock Multiplier. No structural precursor — a genuine black swan with no multi-day buildup detectable in realized data.
Analyst Note
Correct result for an exogenous shock. No realized-data model reliably leads a true black swan. Simultaneous detection via shock mechanism is the honest outcome.
2022Structural Decay
Rate Tightening Cycle
Jun 13, 2022 — Peak drawdown
Lead Time
143 trading days
vs VIX / MOVE (Calibrated)
GSRI leads VIX and MOVE
Primary Driver
Equity-credit co-movement as SPY and HYG fell together throughout H1 2022. Treasury vol elevated throughout — GSRI leading MOVE in a rates-driven crisis is the strongest possible validation result.
Analyst Note
GSRI outperforms VIX and MOVE at calibrated equal-FPR thresholds. Did not fire on Omicron fear in Dec 2021 — only fired when equity-credit stress began Jan 2022.
2011 EU Sovereign Debt Crisis
Structural Decay

Cross-asset correlation and HYG spread widening building through H1 2011. Draghi 'whatever it takes' bottom July 2012.

2023 SVB Collapse
Structural Decay

Equity-credit co-movement deterioration and rising treasury vol building through Jan-Feb 2023 — SVB's duration losses were a direct consequence of the 2022 rate cycle.

2016 Brexit Vote
Exogenous Shock / Early Warning

China episode stress carried forward into Brexit uncertainty. Shock Multiplier activated on the vote-day dislocation.

2010 Flash Crash
Exogenous Shock

Markets -9% in minutes, recovered within hours. Correct result: Shock Multiplier fires; no structural precursor to detect. Tests shock detection, not lead time.

2017 Calm Baseline · Full Year
Peak GSRI 8.7 · Days ≥50: 0 · Days ≥30: 0

251 trading days. VIX averaged 11. Zero days in Monitoring or above. The model is fully quiet during the lowest-volatility year in modern market history.

2019 Calm Baseline · Full Year
Peak GSRI 75.2 (Jan only) · Days ≥50: 7 · All in Jan 2–10

Jan 2–10 shows Q4 2018 correction in the 60-day lookback — correct behavior, not a false positive. By January 11 GSRI dropped to 42. April through December 2019 (SPY +31%, VIX avg 15): clean Normal throughout.

vs VIX · MOVE

VIX measures equity implied volatility. MOVE measures rates market fear. Both spike with the crash, not before it. SRX detects structural decay: slow cross-asset deterioration preceding dislocations by months. The 2013 Taper Tantrum — a pure bond-market event with no equity-credit stress — is a clean non-detection. GSRI leading MOVE in the 2022 rates crisis is the strongest possible result: a multi-asset indicator detecting a rates-driven crisis before the rates fear index itself.

Research Access
Live Platform

Explore the dashboard

Full GSRI monitor with 10-year history, SRS signal decomposition, portfolio stress testing across five scenarios, historical crisis validation across 13 periods, regime intelligence, ML weight comparison, and one-click PDF report generation. Input your own holdings to see portfolio-specific impact.

Methodology Brief

Request the technical brief

For risk managers and portfolio strategists. Covers directional correlation construction, adaptive window mechanics, ML weight optimization, Kalman threshold calibration, and the complete 13-era walk-forward validation record.

Directional correlation signal construction
GSRI component weights & regime-conditional variants
SLSQP weight optimization with 13-era LOOCV
Kalman filter threshold calibration on live signal history
Equity-credit co-movement signal
Adaptive window with consecutive-days activation
13-era walk-forward validation results
SRX

Systemic Risk Analytics — Research Prototype v6.0

shahc.ruhaan@gmail.com

Research prototype only. All metrics and scenarios are for analytical demonstration purposes. Not financial advice. Not a regulated exchange or trading venue.