Risk Watch
Private-credit stress monitoring that separates early market warnings from confirmation in regulatory disclosures and fund flows.
The problem
Private-credit stress rarely arrives as one clean, timely number. Traded markets can move quickly but only provide proxies. Regulatory disclosures can confirm pressure in the vehicles themselves, but they arrive with a reporting lag. Risk Watch was built to keep those two kinds of evidence separate, visible, and useful without overstating what either one can prove.
Two layers of evidence
The early-warning layer monitors public-market and macro credit conditions: high-yield and CCC spreads, bank-credit indicators, listed BDCs, liquid credit, software equities, and leveraged-loan ETFs.
The confirmation layer looks for filed evidence in SEC disclosures, public BDC data sets, N-PORT interval-fund reports, completed tender outcomes, and ICI high-yield fund flows. Together, the layers show whether market concern is unconfirmed, whether filed stress is emerging beneath calm markets, or whether both are deteriorating.
From observations to decisions
Every observation retains its provider, observation time, fetch time, freshness, and proxy status. Changed observations are scored and persisted as snapshots, then streamed to the dashboard.
The scoring model uses fixed component weights. Missing evidence contributes no stress and does not transfer its weight to the signals that remain. Instead, the dashboard lowers evidence coverage and labels partial dimensions explicitly. Operators can inspect raw thresholds, contribution reconciliation, active exceptions, and one-day or one-week changes behind each regime.
Built for operation
Risk Watch is a FastAPI application with live WebSocket updates, replayable snapshots, and alert delivery. It runs locally with SQLite and supports PostgreSQL with TimescaleDB, optional Redis pub/sub, structured logging, Prometheus metrics, provider circuit health, and separate liveness and readiness checks for production deployments.
Collectors poll FRED, Polygon, SEC EDGAR, SEC public BDC and N-PORT data, tender filings, and ICI releases at source-appropriate cadences. Provider failures are isolated, retried, and surfaced without allowing one unavailable feed to stop the collection loop.
Evidence has limits
Public filings can confirm reported NAV marks, non-accruals, fund flows, and tender outcomes, but they remain lagged to their reporting periods. Public prices and bank series are proxies, not direct observations of private-credit portfolios.
Institution-grade monitoring still requires internal administrator or portfolio feeds for current-day NAV, live redemption queues, liquidity terms, borrowing availability, covenant headroom, and vehicles that do not report the needed evidence publicly. Risk Watch makes that boundary explicit rather than filling it with synthetic certainty.