Cross-Border Systemic Risk Monitor
CBSRM reproduces the canonical systemic-risk and financial-stability measures used by central banks and supervisors — in modern, typed, fully-tested Python — under one Protocol, one SHA-256 audit chain, and one reproducibility guarantee. Built around SR 26-2 model-risk-management principles.
Current conditions · live
The same canonical indices CBSRM reproduces — computed from the latest public data, each value dated to its provider's last publication, content-hash stamped, and refreshed automatically every day. Not a backtest; the current read.
Why CBSRM exists
Academic code lives in R, Stata, and MATLAB; commercial systems are black boxes; reference numbers come from vendors that don't agree. None of that clears a model-risk function. CBSRM is methodology-first.
Fragmented implementations, undocumented assumptions, vendor numbers you can't reproduce, and no lineage from input to published figure. A reviewer asks "why is this number what it is?" — and no one can say.
Every number is reproducible from source. Every methodology stage carries its own version. Every lifecycle event is written to a SHA-256 audit chain. One Protocol covers stress indicators and risk-pricing measures alike — adding a method is one file plus one registration line.
The methodology
One pipeline resolves the questions a financial-stability desk actually asks — how stressed is the system, which firms carry the tail, how does shock propagate, and what regime are we in. The same methods now run live in the terminal, alongside a global cross-asset quant cockpit and a turnkey SR 26-2 model-validation binder.
CISS-family stress (US / Euro-Area / UK), STLFSI4, and the OFR Financial Stress Index — quadratic-form aggregation that amplifies in correlated stress.
Expected capital shortfall of a firm conditional on a systemic market decline — who would need recapitalising in a crisis, and by how much.
The system's value-at-risk conditional on an institution in distress, and each firm's marginal expected shortfall — tail dependence, not average correlation.
A pure-numpy DebtRank engine over the interbank exposure network — distress that propagates once through the graph, ranking systemic importance by impact.
Yield-curve recession probit, the Sahm rule, FFR/USD regimes, and a four-state composite that classifies the prevailing macro-financial regime.
Generalized FEVD spillover index measuring cross-asset connectedness — how shocks in one sector transmit to the rest of the system.
The differentiator
CBSRM's L3 Composer assembles every stage into one content-addressed PipelineRecord — the difference between a quant toolbox and a model-governance system.
# Compose one auditable research record for the 2008 crisis window GET /api/pipeline/2008Q4 { "pipeline_id": "crisis-research-flow", "window_id": "2008Q4", "stages": [ /* 7 stages, each version-pinned */ { "stage_id": "score_event", "identifier": "crisis-research-flow:2008Q4:score_event:1.0.0", "output_sha256": "9f12959…" }, … debt_rank · phase_classifier · crisis_dossier · report_renderer ], "hashes": { "output_sha256": "5ee94755…" } ← the join key } # Prove nothing was tampered with after composition POST /api/pipeline/verify → { "ok": true, "mismatches": [] }
See it now
Six desks under one governed roof — live systemic conditions, a global cross-asset quant cockpit, governed crisis dossiers, live capital-shortfall, and a turnkey model-validation binder. Every reading is reproducible, content-addressed, and audit-logged. Start free for 7 days — no card required.
ECB CISS (US/EA/UK), Fed financial stress, yield-curve recession probability, VIX, a real Diebold-Yilmaz cross-border spillover index, HY credit, fund-redemption fragility, ESG transition — each with history & analytics.
World indices, FX, commodities, mega-caps, crypto. Sharpe / Sortino / annualised & EWMA vol / beta / VaR & CVaR / drawdown, plus a candlestick chart with projected levels and a volatility cone.
Reproducible GFC / COVID / SVB dossiers — a system-stress gauge, DebtRank network contagion, regime classification, and a cross-crisis comparison — each a content-addressed PipelineRecord you can verify.
Live SRISK Σ capital-shortfall + per-firm + ΔCoVaR across the major banks, computed with the NYU V-Lab (Brownlees-Engle) and Adrian-Brunnermeier engines from public balance sheets, market cap, and prices.
Model inventory & materiality tiering, conceptual soundness with citations, a materiality-rated assumption register, an SRISK parameter-sensitivity sweep, Monte-Carlo convergence, and an attestation block — so your model-risk function can sign off.
Every reading is content-addressed (SHA-256); every sign-in and access is written to a tamper-evident hash-chained ledger. Rebuild any governed record and verify it reproduces bit-for-bit. The audit trail is the product.
Who it's for
The same code runs on a researcher's laptop and behind a production service — from a single FastAPI deployment to an air-gapped institutional install.
Hedge funds and proprietary trading firms pricing systemic and tail risk with numbers they can reproduce and explain.
Financial-stability and research teams reproducing CISS, SRISK, and CoVaR with point-in-time discipline.
SupTech and RegTech vendors embedding auditable, citable systemic-risk measurement into their stack.
Allocators who want an independent, transparent read on system stress and contagion exposure.
Researchers needing a reproducible, citable reference implementation for SSRN-grade figures.
Engagement
The CBSRM methodology is Apache-2.0 and free to read, run, and cite. What institutions engage us for is delivery: hosted infrastructure, jurisdiction extensions, validation packs, and support.
Request a briefing
A 30-minute walkthrough of the five lenses, the audit chain, and the governed pipeline — with a live, reproducible record for the window of your choice.