How to judge a prediction-market venue
The marketing of this category runs on the same three claims: fair, liquid, transparent. The strongest venues make them and the weakest make them — which is what tells you the words, on their own, carry no information.
That is not an indictment; it is what marketing is. The problem is that most advice on choosing a venue operates at the same altitude — reputation, polish, size, the general feeling that a big number means a safe place. Feelings are how this goes wrong.
Aviation solved this problem culturally, and the solution was not trust. A pilot does not board an aircraft because it looks airworthy. They walk around it before every flight with a written checklist — specific items, fixed order, every time. The walkaround is boring by design. Boring is what checking looks like when the answer matters.
A prediction-market venue deserves a walkaround. Here it is — six questions, in the order they bite: the settlement source, the public settlement record, who quotes the markets, whether the displayed depth is fillable, where user funds sit, and whether anyone answers to a regulator. None of them requires inside access or technical skill. Each is checked by reading: the answer sits in the venue’s published documents, or it is missing.
1 · Is the settlement source named before anyone trades?
Every market is a promise about a future fact, and the promise is only as strong as the rule that will grade it. What clean settlement looks like — the source named in advance, the dispute window stated in hours, the void conditions in plain language — is its own subject, covered in How an honest venue settles a question. The walkaround version is one sentence: find the settlement spec before you commit anything, and if there is no spec to find, none of the next five answers can compensate.
2 · Is the settlement record public — contested cases included?
A venue’s settled questions are its track record, and a track record has three failure modes: hidden, curated — and the subtlest, re-graded, where the rules moved while the record was being written (How an honest venue settles a question). The honest form is the full history — every question, every outcome, every dispute and how it was resolved under the rules as they stood. Public records are what let outsiders grade these prices at all: calibration — whether the 70s actually happen about seven times in ten — is measurable, and researchers have been measuring it for decades, and still are (Page and Clemen 2013; Bürgi, Deng and Whelan 2025, working paper; what the record shows). A venue that publishes its record is volunteering to be graded. A venue that curates it is grading itself.
3 · Who is quoting the markets — and does the venue say so?
Young venues almost always quote their own markets. Someone has to be the counterparty before the crowd arrives, and there are decades of respectable, studied precedent for exactly this arrangement (Market makers, disclosed). The honest architecture is recognizable: the operator’s book ring-fenced from the exchange, disclosed in the rulebook, designed to fade as outside makers arrive. So the question is not whether an operator’s book exists — on a young venue, expect it. The question is whether the venue says so in writing, under named constraints.
4 · Is the depth on the screen fillable?
A book can look deep in two ways: with standing offers that will actually fill you, or with display. From outside, you cannot audit which. What you can ask is what the venue will put its name to. Does it state that every visible quote is a firm commitment, tradable at the shown size? Will it say — even approximately — how much of the visible depth is the operator’s own? Disclosure is the entire test available to a reader, and it is a real test: a venue willing to put those two sentences in writing has bound itself. A venue that will not is showing you display.
5 · What happens to your funds, and under whose rules?
The moment funds arrive at a venue, you hold a claim on that venue — the question is what kind, and written where. Are user balances segregated from the operating company’s accounts, or commingled? Do published terms govern them, or goodwill? And the sharpest version, the one worth reading the documents to answer: if the venue failed tomorrow, who — under which law, in which country — decides what happens to balances? A venue does not need to be large to answer this. It needs to have written the answer down before you asked.
6 · Is anyone accountable to a regulator, anywhere?
This is a fact question, not a virtue question. A complete answer has three parts: the name of the authority, the scope of what it actually supervises, and the rulebook a user can invoke when something goes wrong. The parts matter because they change what the first five answers are made of. Where a regulator with jurisdiction audits segregation, settlement conduct and disclosure, those answers are obligations. Where none does, they are promises — and the difference between an obligation and a promise is who gets to notice when it breaks.
What the six have in common
None of them asks you to judge anyone’s character, and none needs the venue’s cooperation to be asked. Each converts a claim into a piece of paper you can go and find. Fair has a spec. Transparent has a rulebook. Liquid has a name attached.
None of the six is a guarantee — a venue can write a commitment down and break it. But a written commitment can be caught, where an unwritten one can only be doubted, and the sixth question decides who, besides you, is positioned to catch it.
Together they produce a seventh datum, the deepest one: how a venue behaves when asked. The operators most comfortable being measured are the ones who wrote the measurements down before anyone asked. The ones who answer with more claims have answered.
So do the walkaround. Six questions, fixed order, every answer on paper — before funds move, on any venue, every time. Apply it to everyone — including us, when the time comes.
Sources
- Page, L., Clemen, R. T. (2013). “Do Prediction Markets Produce Well-Calibrated Probability Forecasts?” The Economic Journal 123(568): 491–513. academic.oup.com
- Bürgi, C., Deng, W., Whelan, K. (2025). “Makers and Takers: The Economics of the Kalshi Prediction Market.” Working paper, UCD Centre for Economic Research WP25/19. ucd.ie