> ## Documentation Index
> Fetch the complete documentation index at: https://docs.opersa.company/llms.txt
> Use this file to discover all available pages before exploring further.

# The Oracle

> How OperSa prices assets that trade a few times a year.

Pricing a stock is easy. There is a live market and a last trade. Pricing a Nautilus, a pair of Jordans, or a pre-IPO name is the hard problem, because those assets trade rarely and quote nowhere. The OperSa oracle exists to solve that specific problem, and it does it in a way you can audit.

The rule behind it is simple: agents propose, rules decide.

## Two pricing paths

<CardGroup cols={2}>
  <Card title="Live feed" icon="bolt">
    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.
  </Card>

  <Card title="AI valued" icon="brain">
    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.
  </Card>
</CardGroup>

## How a collectible is valued

<Steps>
  <Step title="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.
  </Step>

  <Step title="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.
  </Step>

  <Step title="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.
  </Step>

  <Step title="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.
  </Step>

  <Step title="Decide eligibility">
    Only markets with enough clean, recent evidence become eligible for leverage. Everything else stays spot only. The system is conservative on purpose.
  </Step>

  <Step title="Check the result">
    An independent evaluator reviews the output against the evidence before it is published, a second pair of eyes on every number.
  </Step>
</Steps>

The valuation, the evidence behind it, and the checks are recorded and hashed, so a given price is tied to the exact evidence that produced it.

## Honesty labels

OperSa never dresses up a guess as a fact. Every market tells you how its price was set.

<CardGroup cols={2}>
  <Card title="Live" icon="circle-check">
    The price comes straight from a live on-chain feed. Used for stocks and ETFs.
  </Card>

  <Card title="AI valued" icon="robot">
    The price is an oracle estimate built from real market evidence. Used for collectibles.
  </Card>
</CardGroup>

## 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.

<Card title="The oracle is a registered agent on Virtuals" icon="network-wired" href="/technical/virtuals">
  It runs as an autonomous agent on Virtuals Protocol, the network for on-chain AI agents.
</Card>