Oracle Deviation Threshold
Pronunciation: OR-uh-kul dee-vee-AY-shun THREH-shohld
Definition
An oracle deviation threshold is the configured change that triggers a report update, alert, source rejection, or other oracle action.
Overview
How it works and why it matters
A deviation threshold specifies how much a reported measurement must change before an oracle is prompted to publish again. For a price feed, this is often a percentage difference from the previously published value. It reduces gas consumption in calm markets, but a wide threshold may allow a checkout or lending contract to work with a meaningfully outdated quote. A narrow threshold produces more updates and costs. The decision must consider asset volatility, liquidity, transaction size, and how much price error the application can tolerate.
Technical context
Deviation thresholds reduce unnecessary on-chain updates by publishing when the observed value moves sufficiently from the last report.
A threshold that is too wide leaves applications with stale prices, while one that is too narrow increases fees and can amplify noise.
Integrators should verify the feed’s current threshold and heartbeat on the specific network rather than assuming a global setting.
Practical checks
Translate the percentage threshold into money at the largest order size you accept. A 0.5% price gap on a $20 checkout is small; the same gap on a $200,000 settlement is a $1,000 difference before other costs. Test large jumps that cross the threshold during destination-chain congestion.
Related concepts
Further reading
Example
An ETH/USD feed last published $2,000 and uses a 1% movement trigger. A source observation of $2,010 might not trigger a price update, even though the difference matters to a large order. A checkout accepting only 0.2% pricing error needs its own tighter execution bound.
Comparison
Deviation is about price movement, while heartbeat is about elapsed time. A 1% threshold does not mean all observations are accurate to within 1%, especially during delay or manipulation.
Key Takeaway
Choose thresholds for the application's risk, not just the oracle's update cost.
Sources
- Ethereum Documentation: Oracles — Ethereum Foundation (2026-07-30)
- Chainlink Documentation — Chainlink (2026-07-30)