Oracle Aggregation
Pronunciation: OR-uh-kul ag-rih-GAY-shun
Definition
Oracle aggregation combines multiple data observations or signed reports into one value using rules designed to reduce noise, errors, and manipulation.
Overview
How it works and why it matters
Aggregation turns multiple data observations into one value that a contract can consume. A network may use a median, weighted mean, trimmed mean, or another documented rule. Inputs should be normalized for asset pairs, units, observation times and market liquidity before aggregation. The choice of rule changes how strongly one extreme quotation influences the result. Aggregation also creates dependence on the quality and independence of the contributing sources; a large reporter count cannot compensate for many copies of the same bad input.
Technical context
Oracle aggregation may occur across exchanges within one node, across independent node observations, or at both levels.
Practical checks
Document whether input filtering happens before or after each reporter’s local aggregation. If several reporters first average the same venue and the network later takes a median of their answers, apparent reporter diversity can hide source concentration. Assess time alignment and precision as well as the aggregation formula.
Related concepts
Further reading
Example
Five venues report ETH prices of $2,000, $2,002, $2,001, $2,003, and $2,400. A median returns $2,002, resisting that single extreme observation. If four sources secretly reuse the inflated venue's API, however, the median can still be wrong.
Comparison
Aggregation is not consensus about blockchain transactions; it is a rule for combining external measurements. Median resistance to one outlier is not a blanket guarantee against collusion, stale data, or concentrated liquidity.
Key Takeaway
Examine source independence and calculation method before trusting an aggregated feed.
Sources
- Ethereum Documentation: Oracles — Ethereum Foundation (2026-07-30)
- Chainlink Documentation — Chainlink (2026-07-30)