Oracle Price

The Index Price (also called the Oracle Price) represents the fair spot market price of an asset. It is derived from multiple major centralized exchanges and serves as the foundation for the funding rate calculation and as a reference for the Mark Price.

Data Sources

Each trading pair has a configured set of exchange price sources with assigned weights. For most assets, the default sources and weights are:

Exchange	Weight
Binance	70%
Bybit	20%
OKX	10%

The Oracle Price is published every 3 seconds by the validator node.

Oracle Price = Σ (Price_i × Weight_i)

For the full list of data sources per trading pair, see Supported Assets.

Protection Mechanisms

Several safeguards ensure the Oracle Price remains accurate and resistant to manipulation:

1

Timeout Detection

If an exchange's price feed has not been updated for more than 15 minutes (configurable per pair via MaxDelay), that exchange is excluded from the Oracle Price calculation. Once the feed recovers, it is automatically re-included.

2

Outlier Filtering

If an exchange's reported price deviates from the median by more than the configured threshold (e.g., ±5% for most assets; ±1% for BTC and ETH), the price is automatically capped to the deviation threshold. Once the price returns to a normal range, it is re-included at its actual value.

Example: If the median spot price is 100 and OKX reports 120 (a 20% deviation exceeding the 5% cap), OKX's price is capped to 105 for the calculation.

3

Smoothing Mechanism

To prevent sudden price jumps, the Oracle Price limits the maximum per-update change. For example, with a 0.5% cap per update, if the previous Oracle Price was 99.8, the next Oracle Price is bounded to [99.3, 100.3].

4

Dynamic Weighting

Exchange weights are dynamically adjusted based on two factors:

• Liquidity Weight W(L): Reflects the exchange's trading depth and volume. Measured using the square root of the exchange's 1-minute trading volume, preventing any single exchange from having disproportionate influence.

• Latency Weight W(T): Reflects how quickly the exchange's price data is received. Calculated using an exponential decay function: W(T) = e^(-k × Latency) where k = 0.5. Lower latency results in higher weight.

The final weight for each exchange is proportional to W(L) × W(T), normalized so all weights sum to 100%.