> For the complete documentation index, see [llms.txt](https://docs.prophetmarket.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.prophetmarket.ai/core-concepts/resolution.md).

# Resolution

*The five stage pipeline that turns a real world outcome into a claimable payout.*

Resolution is the most consequential mechanism on Prophet and the one most worth understanding before you commit money. Prophet resolves markets using an ensemble of large language models, and the honest starting point is this: model based resolution is fundamentally non deterministic. The same question, put to the same model twice, can produce different reasoning and, at the edge, different answers.

That is a property of how these models work, not a defect awaiting a patch. Prophet's pipeline is therefore designed around that reality rather than against it. Every stage below exists to make the answer that reaches the chain robust to model variance, not to pretend the variance is absent.

{% hint style="info" %}
**The claim being made**

No single model call is treated as authoritative. The claim is narrower and more defensible: an answer that survives a three hour information buffer, a cross checked ensemble, a retry loop, and a twelve hour on chain review window is far more trustworthy than any individual model call.
{% endhint %}

## The pipeline

1. **A three hour buffer** Prophet waits three hours past the market's resolution time before resolution begins. This gives official sources and news outlets time to publish confirmed outcomes, so the models reason over settled facts rather than early reporting and rumour.
2. **The ensemble runs** Multiple models evaluate the market against its canonical resolution source and their answers are cross checked against each other. Agreement at high confidence means the ensemble is confident. Disagreement means it is not, and the pipeline does not proceed on a coin flip.
3. **The result is posted on chain** Once the ensemble reaches unanimity, the outcome is transmitted to the Conditional Tokens Framework exchange deployed on Polygon.
4. **A twelve hour cooldown** The proposed payout becomes visible but is not yet claimable. This review window is enforced by the smart contract, which means Prophet cannot shorten it, override it, or make an exception to it.
5. **Finalisation and redemption** After the cooldown anyone can call the finalisation function. Payouts flow into the framework and become claimable, and winning shares convert to USDC in your Safe at one dollar each.

## When the models disagree

If the ensemble cannot reach a confident answer on the first attempt, Prophet enters an exponential backoff retry loop. The ensemble pauses, waits, and tries again at growing intervals, which gives it both more published information to work from and more chances to converge on a stable answer. In the worst case this continues for up to two days before the retries stop.

| Case                    | Timing                                                | Outcome                                            |
| ----------------------- | ----------------------------------------------------- | -------------------------------------------------- |
| Confident first attempt | Resolution begins \~3 hours after the resolution time | Outcome posted, then a 12 hour review window       |
| Needs retries           | Up to \~2 days on top of the 3 hour buffer            | Outcome posted once the ensemble converges         |
| Retries exhausted       | After the retry window closes                         | Escalated to manual resolution by the Prophet team |
| Genuinely unresolvable  | After manual review                                   | Market may be voided at 50¢ per share              |

**Best case total**

```
Resolution time  +  3h buffer  +  ensemble run  +  12h review window
```

Approximately 15 hours from the event to a claimable payout.

## After resolution

* Trading in the market stops.
* Winning shares become redeemable for one dollar each once the review window closes.
* Losing shares are worth nothing.
* Unredeemed winning shares remain in your Safe indefinitely and remain redeemable, but they earn nothing from that point on.

## Edge cases the pipeline cannot fully prevent

{% hint style="danger" %}
**Three residual risks, stated plainly**

Models can share the same underlying misinformation and converge on a wrong answer. A market can meet a real world event that fits none of its stated outcomes. And once the on chain review window closes, a finalised payout is irreversible even if reality later contradicts it. Manual resolution and market cancellation exist for these cases, but they cannot undo a finalised outcome.
{% endhint %}

The strongest protection available to you is upstream of all of this: read the resolution criteria before you trade. Most resolution disputes are not really about the outcome, they are about a question that was ambiguous from the start.


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter:

```
GET https://docs.prophetmarket.ai/core-concepts/resolution.md?ask=<question>&goal=<endgoal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal.

The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.
