Divergence Radar

Overview

Divergence Radar is a cross-venue monitoring system that continuously measures price dislocations between Hyperliquid and reference centralized exchanges — Binance, OKX, and Bybit. It detects statistically significant divergences, classifies them by signal type, tracks their lifecycle from detection through resolution, and publishes structured signal events for downstream consumption.

The system operates across the full Hyperliquid perpetual surface, comparing Hyperliquid mark prices against reference venue mid-prices in real time. When a divergence crosses a configurable detection threshold, a signal is emitted with all relevant context: direction, magnitude, asset, venues involved, and estimated opportunity window.

Available on Pro and Enterprise tiers.

Divergence Model

For a given asset at time $t$, the divergence between Hyperliquid and a reference venue is defined as:

$\delta_{t} = \frac{P^{\text{HL}}_{t} - P^{\text{REF}}_{t}}{P^{\text{REF}}_{t}} \times 10{,}000$

where $P^{\text{HL}}_{t}$ is the Hyperliquid mark price at time $t$, $P^{\text{REF}}_{t}$ is the corresponding reference venue mid-price, and $\delta_t$ is expressed in basis points (bps).

A signal fires when the absolute divergence exceeds a configured threshold $\theta$:

$|\delta_t| > \theta$

The sign of $\delta_t$ determines the direction of the dislocation:

• $\delta_t > 0$ : Hyperliquid is expensive relative to the reference venue (hl_expensive).
• $\delta_t < 0$ : Hyperliquid is cheap relative to the reference venue (hl_cheap).

The threshold $\theta$ and the detection window parameters are configurable per asset and per venue pair. Default thresholds are calibrated internally based on historical volatility regimes and are not disclosed publicly. Enterprise clients may request custom threshold configurations for their signal feeds.

Signal Lifecycle

Each signal progresses through a defined lifecycle:

1. Detection — The divergence $|\delta_t|$ crosses the threshold $\theta$. A signal event is emitted with status active.
2. Tracking — While active, the signal is updated with peak divergence, duration, and evolving venue prices.
3. Resolution — The signal is closed when one of the following occurs:
   • Converged — $|\delta_t|$ returns below $\theta$. The opportunity window has closed.
   • Expired — A maximum lifetime is reached without convergence.
   • Invalidated — Market conditions change such that the signal is no longer meaningful (e.g., feed interruption).
4. Outcome — Post-resolution, the signal is annotated with the realized P&L in basis points, assuming frictionless execution at detection and resolution timestamps. This provides a historical performance benchmark for the signal type.

Signal Types

Divergence Radar emits four classes of signals, each capturing a distinct market microstructure phenomenon:

Signal Type	Description	Trigger Condition
basis_divergence	Spot-perpetual basis exceeds normal range on Hyperliquid	The spread between Hyperliquid perp mark price and its spot index deviates beyond a statistically calibrated band
cross_venue_arb	Price divergence between Hyperliquid and a reference CEX	$\|\delta_t\| > \theta$ as defined above, measured pairwise against each reference venue
funding_dislocation	Funding rate deviates from cross-venue equilibrium	Hyperliquid’s implied funding rate diverges from the funding rates observed on reference venues by more than a threshold spread
liquidation_cascade	Cluster of liquidation events on Hyperliquid	Multiple liquidation events within a rolling time window exceed a count threshold, suggesting cascading forced selling or buying

Each signal type uses an independent detection model with its own parameterization. The mathematical structure of the detection models (window lengths, smoothing functions, threshold adaptation logic) is proprietary and not disclosed.

Signal Schema

Every signal event contains the following fields:

Field	Type	Description
signal_id	string	Unique identifier for this signal instance
type	string	One of: basis_divergence, cross_venue_arb, funding_dislocation, liquidation_cascade
coin	string	Asset identifier (e.g., BTC, ETH, SOL)
direction	string	hl_cheap, hl_expensive, or neutral (for cascade signals)
divergence_bps	float	Peak observed divergence in basis points
ref_venue	string	Reference venue (e.g., binance, okx, bybit)
detected_at	string	ISO 8601 timestamp of signal detection
resolved_at	string	ISO 8601 timestamp of resolution (null if active)
status	string	active, converged, expired, or invalidated
outcome	string	Resolution type (populated on close)
pnl_bps	float	Realized P&L estimate in basis points (populated on close)
metadata	object	Additional context specific to the signal type

Delayed Feed API

Signals are published with a 30-minute delay on the delayed feed endpoint. This feed is available to Pro and Enterprise tier subscribers.

Request

GET https://data.aleatoric.systems/v1/signals/delayed?coin=BTC&type=cross_venue_arb&limit=50
Authorization: Bearer YOUR_API_KEY

Query Parameters:

Parameter	Required	Default	Description
coin	No	All coins	Filter by asset identifier
type	No	All types	Filter by signal type
status	No	All	Filter by status: active, converged, expired, invalidated
since	No	24h ago	ISO 8601 start time for the query window
until	No	Now - 30min	ISO 8601 end time (capped at 30 minutes before current time)
limit	No	100	Maximum number of signals to return (max: 1000)
offset	No	0	Pagination offset

Response

{
  "signals": [
    {
      "signal_id": "sig_7f3a2b1c",
      "type": "cross_venue_arb",
      "coin": "BTC",
      "direction": "hl_cheap",
      "divergence_bps": 12.4,
      "ref_venue": "binance",
      "detected_at": "2026-03-13T10:30:00.112Z",
      "resolved_at": "2026-03-13T10:31:42.887Z",
      "status": "converged",
      "outcome": "converged",
      "pnl_bps": 8.1,
      "metadata": {
        "hl_price": 87234.50,
        "ref_price": 87245.33,
        "duration_s": 102.8,
        "peak_divergence_bps": 14.7
      }
    },
    {
      "signal_id": "sig_9e4d8c2f",
      "type": "funding_dislocation",
      "coin": "ETH",
      "direction": "hl_expensive",
      "divergence_bps": 6.8,
      "ref_venue": "okx",
      "detected_at": "2026-03-13T10:28:14.503Z",
      "resolved_at": "2026-03-13T10:35:01.220Z",
      "status": "converged",
      "outcome": "converged",
      "pnl_bps": 4.2,
      "metadata": {
        "hl_funding_bps": 3.2,
        "ref_funding_bps": -3.6,
        "funding_spread_bps": 6.8,
        "duration_s": 406.7
      }
    }
  ],
  "total": 142,
  "offset": 0,
  "limit": 50,
  "delayed_by": "30m"
}

Delayed Feed Rate Limits

The delayed feed endpoint consumes standard API rate limit tokens. See Rate Limits for per-tier quotas.

Real-Time Feed (Enterprise)

Enterprise clients receive signals with zero delay via the Unified Stream. Real-time signals are delivered as SSE events with the divergence_signal event type.

Connection

GET https://data.aleatoric.systems/v1/unified/stream?types=divergence_signal&coins=BTC,ETH
Authorization: Bearer YOUR_ENTERPRISE_API_KEY
Accept: text/event-stream

Event Format

event: divergence_signal
id: evt_a8b3c2d1
data: {"signal_id":"sig_live_4f2a","type":"cross_venue_arb","coin":"BTC","direction":"hl_cheap","divergence_bps":11.2,"ref_venue":"binance","detected_at":"2026-03-13T14:22:01.003Z","status":"active","metadata":{"hl_price":87234.50,"ref_price":87244.26}}

Real-time signals include both active events (at detection) and resolved events (at closure), allowing clients to track the full signal lifecycle as it unfolds.

Historical Replay (Enterprise)

Enterprise clients with historical access can replay past signals for backtesting and analysis:

GET https://data.aleatoric.systems/v1/signals/historical?coin=BTC&start=2026-03-01T00:00:00Z&end=2026-03-13T00:00:00Z
Authorization: Bearer YOUR_ENTERPRISE_API_KEY

Historical replay returns signals without the 30-minute delay and includes full resolution metadata.

Access Tier Matrix

Capability	Free	Basic	Pro	Enterprise
Delayed feed (30-min lag)	—	—	Yes	Yes
Signal type filtering	—	—	Yes	Yes
Real-time signals via Unified Stream	—	—	—	Yes
Historical replay	—	—	—	Yes
Custom threshold configuration	—	—	—	Yes
Dedicated signal webhook delivery	—	—	—	Yes
Per-venue pair breakdowns	—	—	Summary	Full detail
Maximum query window	—	—	7 days	90 days
API rate limit	—	—	Standard Pro	Elevated Enterprise

Use Cases

• Arbitrage monitoring — Track cross-venue mispricings between Hyperliquid and CEXs in real time. Measure opportunity frequency, magnitude, and duration to calibrate execution strategies.
• Funding rate analysis — Identify periods where Hyperliquid funding rates are dislocated from cross-venue consensus, signaling potential mean-reversion or directional opportunities.
• Risk management — Monitor liquidation cascades as an early warning for volatility events. Cascading liquidations often precede sharp price moves.
• Market microstructure research — Analyze signal statistics (frequency, duration, magnitude distributions) to study Hyperliquid’s market quality and its relationship to reference venues.

Next Steps

• Unified Stream — Connect to the real-time event feed
• Rate Limits — Understand throughput constraints
• Error Codes — Handle error responses
• Python SDK — Consume signals programmatically