What a 24-platform prediction market documentation audit can tell you

A prediction market review needs product-specific evidence. A public page can identify a brand without establishing who operates the trading service. An exclusion list can prohibit access in named places without confirming access everywhere else. Two fee percentages may refer to different amounts and cannot automatically be added together.

In the source register frozen on 16 September 2026, our 24-platform catalogue contains 41 source-check records: 38 classified as read and 3 as incomplete. Those are research workflow outcomes, not 38 independently verified financial claims or a pass rate for the platforms. Download the dated source register as JSON to inspect the individual findings.

The sample is the 24 products already selected for our catalogue. It is a purposive sample, not a census or a representative sample of the industry. We followed public operator, documentation and regulator links, used browser-rendered pages where a text reader returned no content, and recorded the question, source URL, check date and limited finding for each record.

Read means we retrieved and assessed the source for the stated question. Incomplete means we saw some content but it did not settle that question. Unreadable means retrieval failed; the latest selected register has no records left in that category after follow-up, but it still contains unresolved questions. No deposits, identity checks, trades, withdrawals or refund claims were performed for this audit.

The terms linked from Futuur contain duplicated passages naming an Anjouan company and a Curaçao company, alongside different governing-law clauses. Reading the first paragraph alone would produce a confident operator label while hiding contradictory text farther down the page.

Our finding is an unresolved contracting entity, not a conclusion that either company lacks a licence or that the platform is closed. The terms also separate real-money restrictions from play-money use. Our Futuur review records those limits instead of treating access to the public site as permission to trade real money.

Opening Our Fees in PredictIt’s trading guide exposes the distinction: the profit charge applies when a share is sold above cost or redeemed at settlement; the withdrawal charge applies separately to the amount withdrawn. Partial sales use purchase order to determine the cost basis. A headline sum would ignore both the different bases and when each charge occurs.

For an arithmetic illustration, a position costing $40 and paying $100 has $60 gross profit. A 10% profit charge takes $6. If the resulting $94 is withdrawn, a 5% withdrawal charge is $4.70, leaving $89.30: $49.30 above the original $40. This example excludes method-specific charges, assumes no other positions or cash flows, and is not an observed transaction. See the complete-trade cost guide for a reusable comparison method.

The Myriad interface terms name restricted trading territories, including Ontario as a province. That wording does not justify marking all of Canada unavailable. Nor does leaving another country unnamed prove that every resident or visitor there can open an account and trade every contract.

The same discipline applies across the register: separate residence, citizenship, physical location and product category. Consult the exact operator terms and any applicable local requirements. Our country controls display the scope and source of a rule; unknown coverage remains unknown. See Myriad and Opinion for the recorded policies.

The downloadable register preserves the selected source-check records and their dates; the linked source pages can change. It does not archive full third-party documents. Count the source records to reproduce the totals, then read each finding beside its source rather than interpreting the status label as a verdict on a platform.

When citing this work, identify PredictHub, the 16 September 2026 snapshot and this methodology page. A useful correction includes the source URL, the relevant passage and its date. Send corrections through our contact page. Assessments remain preliminary; our scoring methodology explains the difference between documentation and account testing.