> For the complete documentation index, see [llms.txt](https://docs.djzs.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.djzs.ai/how-a-verdict-is-made/rulesets.md).

# Rulesets

Two frozen rulesets. Weights, severities and thresholds are code, hashed, and reported in every response under `taxonomy`. The plain names are the ones djzs.ai shows people.

## Prediction markets: `DJZS-PM-v1.0`

Four codes, weights sum to 100, FAIL at 25 (`shared/pm-taxonomy.ts`). All four are live.

| Code     | Name                       | Plain name               | Severity | Weight | Fires when                                                                                                                  | Fix                                                                                                                                          |
| -------- | -------------------------- | ------------------------ | -------- | ------ | --------------------------------------------------------------------------------------------------------------------------- | -------------------------------------------------------------------------------------------------------------------------------------------- |
| DJZS-M01 | NARRATIVE\_RESOLUTION\_GAP | Off the market's terms   | CRITICAL | 30     | The reasoning argues something next to the market's question, not its own date, threshold, definition or settlement source. | Argue the market's own terms: the date it resolves, its threshold or definition, and the source that settles it.                             |
| DJZS-M02 | FALSIFICATION\_ABSENT      | No way to be wrong       | CRITICAL | 30     | Nothing stated would prove the bet wrong before it resolves.                                                                | Write the specific event that would prove you wrong before the market settles.                                                               |
| DJZS-M03 | PROBABILITY\_UNSOURCED     | Odds with no source      | HIGH     | 25     | A probability, stated or implied, has no checkable basis.                                                                   | Name the source, the figure, and the date you read it. A dataset, an official source or your own math counts; a rumor or a feeling does not. |
| DJZS-M04 | CONSENSUS\_NO\_EDGE        | Agreement is not an edge | MEDIUM   | 15     | The stated edge is the consensus or the price itself.                                                                       | Say why the market is wrong. Advisory: it never changes a verdict.                                                                           |

Either CRITICAL code alone is a FAIL. M03 alone weighs 25, which meets the line, so it is a FAIL by itself too. M04 can only change the score and the hash ([Verdicts](https://docs.djzs.ai/how-a-verdict-is-made/verdicts.md)).

Hashes reported as `taxonomy.pm_weights_hash` and `taxonomy.pm_taxonomy_hash`: `0xb4102cd37df7f6bcfdc8d8468296a3bc1e59c41593effc8dbb1cb71922a1bb64` and `0xf7792040e4d30a3736c5b9480fccf5814e02d6392d893da2e5103a9074b7bace`.

## Perp and spot: `DJZS-LF-v1.2`

Eleven codes, weights sum to 200, FAIL at 50 (`shared/audit-schema.ts`, `server/engine-v2/verdict-rule.ts`). Four have engine rules today (`server/engine-v2/deterministic-engine.ts`):

| Code     | Name               | Plain name                  | Severity | Weight | Fires when                                                                            | Fix                                                                    |
| -------- | ------------------ | --------------------------- | -------- | ------ | ------------------------------------------------------------------------------------- | ---------------------------------------------------------------------- |
| DJZS-S01 | CIRCULAR\_LOGIC    | No reason for the trade     | CRITICAL | 30     | A position (leverage or size) is stated and the text shows there is no reason for it. | Say in one sentence why the price should move.                         |
| DJZS-E01 | ORACLE\_UNVERIFIED | Data source not vouched for | HIGH     | 25     | Data is cited but no price source gives it provenance.                                | Name each source, who publishes it, and when you read it.              |
| DJZS-I01 | FOMO\_LOOP         | Chasing the crowd           | MEDIUM   | 16     | The data behind the trade is a social signal.                                         | Replace it with a number you can check, or say why the crowd is wrong. |
| DJZS-X01 | EXECUTION\_UNBOUND | No stop or size limit       | CRITICAL | 15     | A position is stated with no stop and no invalidation condition.                      | Write the stop level and the largest size you will allow.              |

S01 fires only when a reading quotes the text showing no reason is given. An unclear reason is `unknown` and gives a WAIT, never this flag.

Defined, with no rule wired yet: S02 LAYER\_INVERSION (25), S03 DEPENDENCY\_GHOST (18), E02 CONFIDENCE\_INFLATION (18), I02 MISALIGNED\_REWARD (16), I03 DATA\_UNVERIFIED (16), X02 RACE\_CONDITION (9), T01 STALE\_REFERENCE (12). They cannot fire today.

Hashes reported as `taxonomy.weights_hash` and `taxonomy.taxonomy_hash`: `0x083b9ce901ec3defe5da6ef1d83381ff329bf8beddcfa02ebe4e5f75855067b0` and `0x9a7d3e9df2f2bc8855b05bdf604e3354d1d99e8212644f01326ccbb577f69ab7`.

## What the rulesets do not check

Whether your facts are true, whether a quoted price is live, whether the market has already settled, or whether the trade will win. A well-formed but invented source passes the source rule: the rule checks that a basis is stated, not that it is real.


---

# 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 by asking a question.

Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters:

```
GET https://docs.djzs.ai/how-a-verdict-is-made/rulesets.md?ask=<question>&goal=<user_goal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case.

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.
