DSRPT

Real-time stablecoin and crypto stress intelligence

Detect depeg risk, liquidity strain, and contagion regimes before they become treasury losses. For funds, protocols, exchanges, and risk desks.

View Dashboard Request API Access

Live Signal

Updated every 15 minutes. Classification powered by trajectory-shaped features across a 48-hour sliding window.

Asset	Price	Signal Regime	Confidence (?)	Recommended Action
USDC	$1.0000	AMBIGUOUS	28%	Normal Operations
USDT	$0.9991	AMBIGUOUS	31%	Normal Operations
DAI	$0.9999	LIQUIDITY DISLOCATION	80%	Monitor Closely

Confidence = classification certainty. Measures how closely current trajectory features (monotonicity, recovery completeness, severity persistence) match the active regime's rule thresholds. 80%+ = high conviction.

Signal engine online|3 assets monitored|Base Mainnet

USDC Signal

Signal Regimes

Five regimes derived from trajectory geometry — slope, persistence, rebound quality, and cross-venue divergence. Not endpoint rules.

REFLEXIVE COLLAPSE

Monotonic severity path with no recovery signal. Triggers: monotonicity score > 0.55, deterioration run > 25% of window, recovery completeness < 50%. Detected UST 141h before trough.

COLLATERAL SHOCK

Fast spike, fast recovery, bounded persistence. Triggers: recovery completeness > 70%, early/late severity ratio > 2x, persistence < 35%. Detected USDC/SVB 9h before $0.919 trough.

CONTAINED STRESS

Sustained mild elevation, contagion without structural failure. Triggers: max severity 1-12%, persistence > 20%, recovery half-life > 12h. Characteristic of FRAX March 2023.

LIQUIDITY DISLOCATION

High volume, low price impact. Execution risk elevated but not systemic. Triggers: max severity < 3%, persistence < 15%, volume spike ratio > 4x. Venue-specific fragmentation.

Built for

Protocol Treasuries

Monitor stablecoin reserves. Regime alerts trigger rebalancing playbooks before losses materialize.

Trading Desks

Programmatic risk signals for position sizing, hedging, and exposure management across stablecoin holdings.

Exchanges

Cross-venue spread monitoring, deposit/withdrawal circuit breakers, collateral haircut automation.

Risk Managers

Decision support: monitor, reduce exposure, pause deposits, tighten haircuts, escalate to manual review.

Signals API

Machine-readable risk signals. Same data as the dashboard, structured for integration.

GET/api/v1/signals/assets?symbol=USDC

{
  "asset": "USDC",
  "price": 0.9998,
  "regime": "ambiguous",
  "regime_id": 0,
  "confidence": 0.28,
  "escalation": 0,
  "peg_dev_bps": 2,
  "max_severity": 0.0009,
  "partial_scores": {
    "reflexive_collapse": 0.47,
    "collateral_shock": 0.38,
    "contained_stress": 0.58,
    "liquidity_dislocation": 0.46
  },
  "recent": [ ... ]
}

GET /api/v1/signals/marketComposite market signal + all assets

GET /api/v1/signals/assets?symbol=USDCSingle asset with partial scores + 24h history

GET /api/v1/alertsRegime transitions, warnings, confidence shifts

GET /api/v1/history?symbol=USDC&range=30dTime series with event markers for charting

Methodology

Validated against real depeg events. The classifier uses trajectory features — not where the price ended, but how it got there.

UST / May 2022

Reflexive collapse. First alert at $0.984, 141h before terminal trough at $0.018. Key features: monotonicity score 0.72, deterioration run 84h (60% of window), early/late ratio 0.21, recovery completeness 0.03. No false recovery signal.

USDC / March 2023

Collateral shock. First alert at $0.985, 9h before trough at $0.919. Key features: recovery completeness 0.89, early/late ratio 3.2, peak recovery asymmetry 0.4, persistence 0.28. Correctly classified as bounded, not structural.

Trajectory Features (v2 classifier)

Monotonicity score — fraction of timesteps where severity is non-decreasing. High = one-way deterioration.
Recovery completeness — how much of peak severity has resolved. Low + high monotonicity = collapse.
Early/late ratio — mean severity in first 25% vs last 25% of window. Below 0.40 = worsening conditions.
Deterioration run — longest consecutive increasing-severity window in hours. Long runs = structural failure.
Abandonment signal — gap between raw and adjusted recovery. Detects volume collapse masking terminal severity.

Full Research Paper

On-chain Risk Products

Parametric depeg protection powered by the signal engine. 6 contracts deployed on Base Mainnet.
Pilot access only — contact for details.

Request Access

Get API keys, dashboard access, or schedule a demo.