Price feeds and oracle infrastructure
We provide price feeds for digital assets and tokenized real-world assets, including assets that do not trade continuously and therefore cannot be priced by reading a tape. Feed integrity is monitored by the same surveillance that runs on the books we quote.
Pricing what does not trade
A price feed for a liquid token is an aggregation problem: take the venues where it trades, weight them sensibly, reject the outliers, and publish. The engineering is well understood and the failure modes are known.
Tokenized real-world assets break that model, because the underlying may trade rarely or never on a public venue. The price then has to be produced by a stated method — from the underlying market where one exists, from comparable instruments, from a valuation schedule, or from the primary issuance level — and the method matters more than the number. A feed that publishes a price without disclosing how it was derived is asking every downstream consumer to trust it blindly.
Because we make markets, we see manipulation attempts from the inside. That is the direct input into what a manipulation-resistant feed needs to reject.
What is a price oracle?
A service that brings a price on-chain so that contracts can act on it. Anything that liquidates a position, marks collateral, settles a derivative or computes a redemption depends on one. Because those actions are automatic and irreversible, the oracle is usually the most attractive thing in the system to attack.
How do you price an asset that does not trade continuously?
By a stated method rather than by observation. Depending on the instrument that means referencing the underlying market, a basket of comparables, a scheduled valuation, or the primary issuance level, with defined behaviour for how the published price moves between observations. The method is disclosed, because a consumer of the feed needs to know what the number represents.
What makes a price feed manipulation-resistant?
Aggregating across venues so no single book sets the price, weighting by real liquidity rather than reported volume, using time-weighting so an instantaneous print cannot move the published value, rejecting deviations beyond a threshold, and defining what the feed does when sources disagree or go stale. The last one is where feeds usually fail, because staleness is less obvious than a bad tick.
How is feed integrity monitored?
By the same surveillance that runs on the books we quote. A manipulation attempt aimed at an oracle shows up first as anomalous activity on a source venue, which is exactly what the surveillance is built to detect. The feed and the monitoring are not separate products bolted together.
What you receive
- Published feed with documented derivation method and update frequency
- Source venue list with weighting and liquidity basis
- Defined deviation, staleness and fallback behaviour
- Monitoring of source venues for manipulation targeting the feed
How we engage
Engagements begin with a conversation about the instrument, the venues involved and what you are trying to achieve, followed by a written scope. Terms, including how we are compensated, are disclosed in summary before anything is signed.
We do not publish pricing, because scope varies enough that any number on a page would be wrong. Ask and we will tell you.
An oracle is a claim about a price. We are in the market that price comes from, which means we can be held to it rather than merely publishing it.