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

# Customer fair values

> Future private estimates, separate from market observations and structural payouts.

<Note>Future extension. These operations are proposals beyond the initial offering; storage and probability-weighted analytics are not available.</Note>

## Bring your own estimates

The proposed interface would store customer-supplied outcome probabilities or expected payouts in an authorized book/strategy scope. Values remain private to the customer. They are not Rook forecasts and do not change settlement rules.

| Operation | Intended behavior |
| - | - |
| [`POST /fair-values:batchUpsert`](/api-reference/batch-upsert-fair-values) | Atomic batch upsert of explicitly supplied estimates |
| [`GET /fair-values`](/api-reference/list-fair-values) | Filtered reads in an authorized `scope_id` |
| [`DELETE /fair-values/{id}`](/api-reference/delete-fair-value) | Delete a private estimate, with optional revision precondition |

Records retain the outcome, market version, estimate type/value, source, observation time, optional validity deadline and server revision. A probability must be between zero and one with `probability_event: FULL_PAYOUT`; expected payout requires the matching currency/unit convention.

Expired or superseded estimates remain explicitly stale. A settlement amendment requires revalidation. Duplicate batch keys, invalid records or revision conflicts reject the whole upsert batch.

## Marginals are not a joint forecast

If a desk supplies `P(A)=0.60` and `P(B)=0.50`, those inputs alone do not determine the probability of an A-AND-B ticket. Without additional restrictions its probability can range from 0.10 to 0.50. Multiplying the marginals assumes independence.

Settlement relationships can tighten bounds or reveal inconsistent inputs. They cannot establish calibration or silently create a unique joint distribution.

## Probability-weighted analytics

A future extension could accept normalized weights over a version-pinned supported joint scenario scope. Expected payout would be the weighted sum of exact scenario payouts. Such weighting would not remove zero-weight scenarios from structural worst-case analysis.

Saved marginal estimates are not substitutes for joint weights. Initial evaluation requests do not accept these future probability fields. Customer simulation-sample mapping and Rook-generated forecasts remain separate product decisions.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.