Buyer's guide ยท Prediction Stack

Prediction market analytics source code: Kalshi spreads and Polymarket on-chain data

Prediction Stack is the source code for read-only prediction market analytics that you host yourself as static pages. It checks single-venue Kalshi spreads with every fee counted, runs a strict cross-venue matcher between Kalshi and Polymarket, and decodes Polymarket fills, fees and builder-code attribution from Polygon logs. A GitHub Actions job fetches the data and commits it as JSON, so there is no server in the visitor's path. It places no orders and does not need exchange keys.

What you get

  • Spread Scanner

    A single-venue Kalshi check with the Kalshi fee counted on every leg and the spread annualised over the days capital is locked. Rows that are research-only or lack order-book depth are flagged, and a zero is shown as zero.

  • Cross-venue matcher

    Kalshi to Polymarket matching with vetoes for settlement, year, subject, source, predicate, threshold and party. Only pairs a human has confirmed are priced. It is marked partial in the product files.

  • Prediction Terminal

    Read-only. Polymarket fills, fees and builder-code attribution decoded from Polygon RPC logs, using your own RPC or the built-in public fallbacks.

  • Refresh job

    A GitHub Actions workflow with an integrity gate. It runs on manual dispatch, and the schedule is commented out.

  • Operator admin

    Built on the MIKODES Admin Kit. Settings, status checks, audit, a pair review queue, legal links, venue toggles and a category filter. The admin writes config/public.json and runs only on 127.0.0.1.

  • Pages and docs

    A hub page, a shared design system with themes, and more than 15 static documentation pages with search.

Who runs it

The product is for people who publish or study prediction market data. It is analytics only.

  • A data or analytics publisher

    Host a branded site of prediction market data on any static host. You set the name, theme, legal links, announcement and categories.

  • A research team

    Use the fee model and the matcher's veto rules as a documented base for your own work, with the Kalshi fee schedule included in the data.

  • A developer

    Build on the tested Kalshi and Polymarket fee, matching and Polygon decoding code.

  • Your responsibility as operator

    Prediction markets are restricted in many jurisdictions. The docs say the operator is responsible for the rules where they publish, and the product never routes around a network or country block.

What it needs to run

  • A GitHub repository with Actions enabled and write access, to run the refresh job. Actions minutes are billed on private repositories
  • Any static host for the repository root, and a domain. The admin folder must never be deployed
  • Node.js 20.12 or newer for the operator admin; its dependencies include a native better-sqlite3 build
  • Recommended: your own Polygon RPC URL as an Actions secret
  • No Kalshi or Polymarket account or API key. Polymarket may be blocked in some countries

What it does not do

  • It does not place, sign or manage orders. It is read-only analytics.
  • It does not promise any found spread. The data committed in the package shows 0 positive spreads and 0 cross-venue matches.
  • Polymarket fee parameters are not verified. The formula is checked against the official SDK, and unverified parameters display as unknown instead of a guess.
  • The builder-fee view is a planner, not a record of income.
  • There is no bot in this version.
  • It does not give legal clearance for your country, and access where a venue is blocked is not included.
  • Hosting, Actions minutes and a private RPC plan are not included.

Questions

Does it place trades?

No. There is no order placement or signing of any kind. It reads public data.

Why do some fees show as unknown?

The Polymarket fee formula is checked against the official SDK, but the live parameters have not been verified. The product shows unknown instead of inventing a number.

Does it find spreads?

That depends on live data. The data committed in the package shows no positive spreads and no cross-venue matches, and the product shows honest zeros instead of filling the page.

Do I need a server?

No. The data job runs on GitHub Actions and commits JSON, and the pages are static. The admin runs only on your own machine and is never deployed.

Does it work in every country?

No. Prediction markets are restricted in many places, Polymarket may be blocked, and the operator is responsible for the legal side.

How much does Prediction Stack cost, and what does the licence allow?

$99, one payment, no subscription. One licence, no tiers: any number of projects and clients, a paid SaaS, every line changed. The only limit is reselling or publishing the source code itself. The services you connect bill you directly. See the licence.

Is setup included?

Yes. We install Prediction Stack for you and put in your logo, colours and texts. Bigger changes or new features get a fixed price before we start.

What if it does not work?

Write to us within 14 days. We fix it, usually the same day. If we cannot fix it within 7 days, you get your money back. See the refund policy.