Trust
How we score platforms
Every number on this site comes from the same model, assessed by our editorial team and stored with its revision history. This page is the model. If you disagree with a weight, you can see exactly which one moved a platform and by how much.
We publish three separate numbers and never blend them: an editorial score out of ten, a user score out of five, and a popularity rank derived from search and traffic data. Popularity is a measure of attention, not of quality, and is labelled as such wherever it appears.
The nine criteria
Each criterion is scored 0–10 with a written justification. The weighted sum is the Editor Score, out of 100 points of weight in total. Weights were set once, in public, and changing one requires re-scoring every platform in the catalogue.
Liquidity & volumes
20%
Spreads on top markets, order book depth, monthly volumeFees & costs
15%
Trading fee, withdrawal fee, spread, gas for on-chainRegulation & trust
15%
Licensing, incident history, fund segregationMarket variety
12%
Categories, active market count, listing speed for new eventsDeposits & withdrawals
12%
Methods, speed, minimums, KYC triggersUX & mobile
10%
Apps, execution speed, interfaceResolution
8%
Oracle transparency, history of disputed resolutionsBonuses & promos
4%
Availability and fairness of termsSupport
4%
Channels, response time (measured in-house)
Unrated criteria. A criterion we have not assessed yet is excluded rather than scored zero, and its weight is redistributed proportionally across the ones we have rated. A half-finished review therefore reads as a score in the right range rather than being dragged toward zero by absent data.
Preliminary scores. Below 80% of the total weight, the score is published with a “preliminary” label and the review says which criteria are still open.
Payment cannot move a score. The function that computes the Editor Score receives the nine ratings and nothing else — it is never passed the platform record, so it cannot read whether we earn a commission from that platform. Commercial status decides whether a placement is labelled “Ad”; it has no path into the arithmetic.
How a review is researched
- Identify the product and operatorDistinguish the exchange, broker, protocol and play-money product before comparing them.
- Read primary sourcesUse published fee schedules, terms, rulebooks and regulator records. Record the source and date in the review.
- Assess what the evidence supportsExplain each scored criterion. Leave missing evidence unrated and disclose estimates and limitations.
- Separate research from account testingOnly a dated test record supports a claim about our own deposit, trade or withdrawal. Reading a fee schedule is not a funded-account test.
- Publish coverage and limitationsShow the share of the scoring model assessed and mark incomplete assessments as preliminary.
- Record material changesReassess affected criteria when evidence changes. Score revisions are logged; a publication date is not a guarantee that every fact remains current.
How we make money, plainly
- →Affiliate commission when a reader opens an account through our link. A commission is possible only where a partner arrangement exists.
- →Paid features carry an “Ad” label. They do not change Editor Scores or the order of editorial rankings.
- →Display advertising sold directly, kept out of the ranked tables.
- →Platforms with no affiliate programme are reviewed on identical terms; their links still route through /go/ so we can count clicks.
Recent score changes
- Hedgehog Markets — Editor Score 1.7, first published score Preliminary
- Space — Editor Score 2.6, first published score Preliminary
- Probable — Editor Score 1.8, first published score Preliminary
- Xmarket — Editor Score 1.8, first published score Preliminary
- Trueo — Editor Score 5.2, first published score Preliminary
- Opinion — Editor Score 5.7, first published score Preliminary
Our policies
Found a factual error? Write to bestpredictionmarkets@proton.me. We review reported errors and note material corrections on the page. Platforms may request a re-test after a material product change; a re-test never guarantees a higher score.