# Slinky Consensus

Public models are indexed, scored, and rewarded by **Slinky Consensus**.\
\
Creators may keep a policy private or publish it; published policies enter a transparent, on-chain ranking system.

***

## Publication Modes

<table><thead><tr><th width="93.7933349609375">Mode</th><th>Registry entry</th><th>Access control</th><th>Protocol fee</th></tr></thead><tbody><tr><td><strong>Private</strong></td><td>none</td><td>Owner API key only</td><td>0</td></tr><tr><td><strong>Public</strong></td><td><p><code>{modelCID, configCID,</code></p><p> <code>owner, fee_bps}</code></p></td><td>Open signal stream</td><td>optional listener micro-fee¹</td></tr></tbody></table>

> ¹ Micro-fees are routed to the treasury and **do not** fund the creator/curator payout discussed below.

***

## Slinky Score

For a public model *i* at time *t*

<figure><img src="https://4263197242-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FtVM0qTwW1SmwvBaQwlXq%2Fuploads%2FPjyEK8OBJogUnGO7Aikr%2Fimage.png?alt=media&#x26;token=8c9289ef-eac6-4b92-965d-35fabf98eb26" alt=""><figcaption></figcaption></figure>

| Symbol            | Definition                                | Default                       |
| ----------------- | ----------------------------------------- | ----------------------------- |
| E<sub>i​</sub>(t) | equity value after latest verified signal | —                             |
| τ                 | trailing window                           | 7 days                        |
| R<sub>i</sub>​(t) | net return over trailing window           | —                             |
| Σ<sub>i</sub>​(t) | $SLINKY currently staked on model *i*     | dynamic                       |
| A<sub>i</sub>​(t) | audit-pass flag (0 or 1)                  | 1 when ≥ 3 identical receipts |

***

## Reward Emissions

The protocol mints **E** fresh $SLINKY **every seven-day epoch** and allocates it by score:

<figure><img src="https://4263197242-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FtVM0qTwW1SmwvBaQwlXq%2Fuploads%2Ft6HUYJ4KOH1YRnnotkC2%2Fimage.png?alt=media&#x26;token=f9298a7e-612b-4688-8a59-57b5a10870e2" alt=""><figcaption></figcaption></figure>

<table><thead><tr><th width="180.25">Recipient</th><th>Share of  epochPool ( 𝑖 )  </th></tr></thead><tbody><tr><td>Creator</td><td>50 %</td></tr><tr><td>Curators</td><td>50 %, split pro-rata by stake weight and lock-time</td></tr></tbody></table>

Early stakers gain leverage because their weight matters most while Σ<sub>𝑖</sub>​ is still small.

***

## Forward Audit

* Signal must arrive ≤ 5 s after candle close.
* Auditors replay signals with canonical close price; identical equity-curve hash ⇒ receipt.
* Quorum (≥ 3 identical receipts) flips A<sub>i</sub>​ to 1, credits auditors from a separate inflation slice, and unlocks the model’s share of the epoch pool.

***

## Uniqueness Guard

&#x20;Before `publish()`, the Registry hashes the weight tensor: ***H*****&#x20;= SHA256(weights).**\
If ***H*** already exists, the transaction reverts, eliminating duplicate uploads.