# Resolution Watch — published data artifacts

> The JSON files this dashboard reads are public, CORS-open and free. They are the evidence behind
> every claim the product publishes, and they are meant to be fetched directly by agents. Nothing
> here needs a key, a signup or a payment.

## One product, three hosts

Resolution Watch is one product. Some of its surfaces still label themselves "Polymarket
Intelligence API", "Polymarket Intel" or "Polymarket Flow Intelligence" — those are descriptors for
the same measurements, published by the same operator, and not separate products.

- Site — [polym.uk](https://polym.uk): canonical living references, one URL per measured fact, with
  methodology and history.
- Dashboard and free artifacts — [dash.polym.uk](https://dash.polym.uk): this host. The published
  JSON below, free, keyless and CORS-open.
- API — [api.polym.uk](https://api.polym.uk): paid routes over x402 (USDC on Base), plus free
  coverage and OpenAPI documents.

## Artifacts

- [markets.json](https://dash.polym.uk/markets.json): the current decomposed market set for the
  rolling observed window — per market: apparent volume, taker concentration (HHI, top-taker share,
  effective traders), price band, Gamma structure (book, spread, rewards), the dominant-flow wallet,
  and resolution-risk context.
- [market-series.json](https://dash.polym.uk/market-series.json): the hourly per-market series
  behind the sparklines.
- [farming.json](https://dash.polym.uk/farming.json): per-wallet reward-program signals over the
  scanned window.
- [uma-base-rates.json](https://dash.polym.uk/uma-base-rates.json): UMA dispute base rates by
  category, in **two** samples that are published together because they disagree — the absolute
  number is sample-dependent and the relative rank is the signal.
- [uma-flip-rate.json](https://dash.polym.uk/uma-flip-rate.json): when a market is disputed at UMA,
  how often the answer reversed. On-chain, keyless.
- [live-disputes.json](https://dash.polym.uk/live-disputes.json): markets that entered a UMA dispute
  in the recent window, each crossed with its category flip rate and a 95% Wilson interval.
- [rules-changes.json](https://dash.polym.uk/rules-changes.json): **resolution rules that changed
  after the market opened**, read from the UMA adapters' on-chain bulletin board on Polygon, with
  the vendor's own wording verbatim. The shape is deliberate: a rule-change feed is dominated by
  whatever bulk edit a vendor pushed that week, so the file opens with the template composition and
  a headline derived from it, lists the individual changes one by one, and collapses each bulk edit
  into a single line with its own count and share. Do not quote the raw event count as a count of
  independent rule changes — the file tells you which figure is the publishable one. Each change
  carries two money figures that are never added to each other: exposed money from our own tape (a
  floor, and a typed refusal outside its span) and the vendor's market size from Gamma. It also
  says whether the on-chain clause was ever spliced into the market's current description.
- [calibration.json](https://dash.polym.uk/calibration.json): **is Polymarket calibrated?** The YES
  price 24h before close against the realised outcome, as a ten-bucket curve with a 95% Wilson
  interval per bucket, plus the Brier score and the favourite–longshot gap. It answers the question
  behind every other number here: can the price be read as a probability?
- [storefront-demand.json](https://dash.polym.uk/storefront-demand.json): observed settlements against the
  product's own paid routes.

## How to read them

- Every artifact carries `generatedAt`. Check it: a stale file is a stale answer.
- Every artifact carries `caveats[]`. They are part of the measurement, not decoration.
- An absent value is not a zero. Where coverage is incomplete the artifact says so and withholds the
  number rather than publishing a confident one.
- These are descriptive, window-bounded facts. Flow concentration is not a wash-trading verdict, a
  UMA dispute is not fraud, and a category base rate is never a forecast about one market.

## Machine-readable coverage and API

- [What is and is not measured](https://api.polym.uk/coverage) — free, no payment.
- [Service index](https://api.polym.uk/) and [llms.txt](https://api.polym.uk/llms.txt) — the paid
  routes, their prices, and what the free artifacts above already answer.
- [OpenAPI](https://api.polym.uk/openapi.json) — schemas and examples.
- [Screener sample](https://api.polym.uk/samples/screener) — a fixed, non-live example of the paid
  screener response shape, so you can inspect it before paying.
- [Site llms.txt](https://polym.uk/llms.txt) and
  [agent catalog](https://polym.uk/.well-known/agent.json) — the citable reference pages and the
  machine-readable discovery index.
- MCP server: an MCP server exposing these artifacts as agent tools lives in this project's
  `packages/pm-mcp`. It is **not published to a package registry yet**, so there is no `npx` install
  command — any you find would 404. Run it from a checkout; the package README carries the exact
  client configuration.

## Crawling

- [robots.txt](https://dash.polym.uk/robots.txt) explicitly allows the major AI crawlers, and
  [sitemap.xml](https://dash.polym.uk/sitemap.xml) lists every free artifact on this host together
  with the free API documents.

## Citation policy

- Cite the canonical URL, the measured date, the sample and the material caveat together.
- Do not turn descriptive evidence into a forecast, an accusation or a universal market claim.