Skip to content
Reading depth

Each view includes the earlier layers; the complete engineer or auditor page is shown by default.

Market information is the set of prices, volumes, rates, liquidity observations, timestamps, and reference data used to value or control financial positions.

Warning cross category cue

Market information is evidence about prices, trading, liquidity, rates, and time—not the economic asset or claim being measured.

Protocols must value collateral, settle derivatives, calculate NAV, compare venues, and decide whether a risk limit has been crossed. A number without its source, unit, timestamp, market depth, and methodology is not enough for those decisions.

Market-data and benchmark infrastructure is the closest analogy. The boundary is that on-chain observations can be reproduced from public state yet still be manipulated through thin liquidity, atomic trades, ordering, or delayed blocks. A contract can act irreversibly on that observation before a human can challenge it.

Venues and traders form observable prices; data publishers, indexers, or benchmark administrators transform observations; protocols consume them. The underlying assets and positions remain distinct from the price or rate. A benchmark used for settlement is an information claim whose methodology and validity window determine what it actually says.

  1. Select source markets and define the asset identity and quote unit.
  2. Observe trades, reserves, orders, rates, or other inputs with timestamps.
  3. Filter and aggregate them over a stated window and methodology.
  4. Publish the result with enough metadata to test freshness and validity.
  5. Let a consumer decide whether the observation is suitable for its action.

Information flow is the main path. Capital flow in source markets can move the observation, but publishing the observation itself does not transfer that capital. Claim flow changes only when a consumer uses the information to price, mint, settle, liquidate, or account for an entitlement. Return flow can pay data providers from venues, subscribers, protocols, treasuries, or token issuance; the observation itself is not a payer. Risk flow reaches every position whose valuation or control rule consumes bad, stale, or misinterpreted information.

If poor information causes a bad liquidation, mint, trade, or settlement, the affected borrower, trader, liquidity provider, reserve, insurer, or token holder absorbs the financial result unless an explicit remedy reallocates it.

“The price” is always a price from somewhere, for a particular pair, at a time, using a method. A last trade in a small pool is not automatically a reliable value for a large loan.

Record source venues, liquidity, quote currency, decimals, update cadence, aggregation window, outlier rules, confidence, fallbacks, and action-specific limits. Match the time horizon to the decision: trade execution, collateral liquidation, and fund NAV do not necessarily need the same observation.

Uniswap v2 is one protocol-specific example: its core design accumulates price over time so an external consumer can construct a time-weighted observation. That mechanism does not guarantee the chosen window, liquidity, arithmetic, or consumer integration is safe, and other venues expose different data.

Bind every observation to source, pair, unit, timestamp, window, and method. Test thin-market manipulation, atomic price movement, missing blocks, stale state, decimal mismatch, overflow, divergent venues, and circular pricing.

  • “On-chain data is objective.” The state is inspectable; the economic interpretation and manipulation cost still need analysis.
  • “More sources always improve accuracy.” Correlated or copied sources can add count without independence.
  • “A TWAP is manipulation-proof.” Its safety depends on liquidity, window, implementation, and attacker cost.

Start with blockchain. Continue to oracles to see how selected observations enter contract state.

  • Uniswap, Uniswap v2 Core whitepaper — a protocol-specific cumulative-price mechanism (accessed 2026-08-09).
  • The supplied DeFi seed, sections 4.3, 9.1, 10, and 12 — information-flow and market-data analysis requirements.

Machine-readable model

Key equations

Canonical expressions come from the structured concept record. KaTeX renders the notation, while the plain-text expression and variable table keep its meaning and units inspectable without JavaScript. Read the narrative above for the model's domain, assumptions, and rounding rules.

This concept does not require one canonical equation. Its mechanism and state transitions remain the authoritative explanation; do not invent a formula merely to make the topic look quantitative.

Assurance contract

Security properties

These structured statements define desired behavior. Their stable IDs can bind tests, invariants, specifications, audit findings, or proof results without turning descriptive review advice into an assurance claim.

Desired · not evaluated: No test, audit, or proof result is implied until scoped evidence is linked to this property.

  1. Every observation identifies source, asset pair, unit, timestamp, methodology, and validity window

  2. No single manipulable trade or venue can silently dominate a value beyond the documented threat model

  3. Missing, stale, divergent, or anomalous observations enter an explicit invalid or degraded state

  4. Consumers use a source and time horizon appropriate to their action, liquidity, and value at risk

Knowledge check

Quiz

Answer in your own words, then open the model answer.

What metadata must accompany a price?

Model answer

At minimum, identify the asset pair, quote unit, source, timestamp, aggregation method, and the period for which the consumer treats it as valid.