How It Works

Rubicon operates as a HIP-3 perpetual deployment on Hyperliquid. Here's how the pieces fit together.

Architecture Overview

┌──────────────────┐     ┌──────────────────┐     ┌──────────────────┐
│  PRICE SOURCES   │────▶│   RUBICON ORACLE │────▶│   HYPERLIQUID    │
├──────────────────┤     ├──────────────────┤     ├──────────────────┤
│ Polygon REST     │     │ OracleSystem     │     │ ETF DEX          │
│ Yahoo Finance    │     │  (3s loop)       │     │  - SOXX-PERP     │
│                  │     │                  │     │  - KODEX-PERP    │
│                  │     │ Submitter        │     │                  │
│                  │     │  - setOracle()   │     │ Mark price calc  │
│                  │     │  - validation    │     │ Funding rates    │
│                  │     │  - rate limit    │     │ Liquidations     │
└──────────────────┘     └──────────────────┘     └──────────────────┘

Key Components

1. Price Sources

We fetch ETF prices from professional data providers:

• Primary: Polygon.io REST API — Real-time equity data

• Secondary: Yahoo Finance — Backup when Polygon unavailable

• Fallback: Cached prices (60s max staleness)

Importantly, we fetch the actual ETF price—not a basket of underlying stocks. The ETF provider handles all rebalancing.

2. Oracle System

Our oracle runs a continuous loop:

1. Fetch current ETF prices from data sources

2. Validate prices are within expected bounds

3. Smooth via EMA for mark price suggestions

4. Submit to Hyperliquid every 3 seconds

The oracle publishes three price types:

• oraclePxs — Authoritative price for liquidations/funding

• markPxs — EMA-smoothed suggestion for mark price

• externalPerpPxs — Circuit breaker reference

3. HIP-3 on Hyperliquid

HIP-3 is Hyperliquid's protocol for deploying custom perpetuals:

• registerAsset2() — Deploy a new perpetual contract

• setOracle() — Publish oracle prices

• Order book — Hyperliquid's matching engine handles all trades

You trade on Hyperliquid's infrastructure with our oracle prices.

Trade Execution Flow

1. Trader submits order via Hyperliquid
           │
           ▼
2. Order enters Hyperliquid order book
           │
           ▼
3. Matching engine executes against resting orders
           │
           ▼
4. Position created/modified
           │
           ▼
5. PnL calculated using mark price
   (influenced by our oracle submissions)
           │
           ▼
6. Liquidations triggered if margin insufficient
   (uses oracle price)

Market Hours Behavior

Market State	Oracle Behavior	Trading
Open	Live price updates every 3s	Full functionality
Closed	Price frozen at last close	Positions hold, no new liquidations from price moves
Pre/After Market	Live updates from extended hours data	Full functionality

Reliability Guarantees

• State persistence — Oracle state survives restarts

• Fallback chain — Multiple data sources prevent outages

• Degraded mode — Alerts trigger after 3 consecutive errors

• EMA smoothing — Prevents mark price manipulation

See Oracle Infrastructure for detailed technical documentation.