Two pricing paths
Live feed
Stocks and ETFs read a live on-chain price feed. There is no estimation. The market tracks the real price directly, and the panel is labeled as live.
AI valued
Collectibles are valued by an autonomous agent that sources real market evidence. The panel is labeled as AI valued, so you always know which path set the number.
How a collectible is valued
1
Gather evidence
An autonomous valuation agent collects recent comparable sales and live listings for the asset, the raw signal a human appraiser would look for.
2
Clean the data
The engine drops observations that are stale or that sit too far from the pack. A single strange sale cannot drag the number around.
3
Weigh what is left
The surviving evidence is combined with a weighted, trimmed median, which leans on the reliable middle of the data and ignores the extremes.
4
Score the confidence
The engine measures how much the evidence agrees with itself and how fresh it is, then attaches a confidence score. Thin or noisy evidence lowers the score.
5
Decide eligibility
Only markets with enough clean, recent evidence become eligible for leverage. Everything else stays spot only. The system is conservative on purpose.
6
Check the result
An independent evaluator reviews the output against the evidence before it is published, a second pair of eyes on every number.
Honesty labels
OperSa never dresses up a guess as a fact. Every market tells you how its price was set.Live
The price comes straight from a live on-chain feed. Used for stocks and ETFs.
AI valued
The price is an oracle estimate built from real market evidence. Used for collectibles.
Why this matters
Anyone can print a number. The point of the oracle is that the number is defensible. It is built from real evidence, it is cleaned by fixed rules rather than opinion, it carries a confidence score, and it is checked before it ships. That is what makes a market on a rare asset trustworthy instead of arbitrary.The oracle is a registered agent on Virtuals
It runs as an autonomous agent on Virtuals Protocol, the network for on-chain AI agents.
