How to run prediction market research

A quoted price on an event contract is easy to screenshot and hard to use well. The same headline can exist on several venues with different wording, different settlement sources and different close dates, and a price that looks like a forecast is also shaped by who is trading, how much size is available and what fees apply. The steps below separate the number from the claim you want to make about it.

1. Write the question before you look at a price

Start with the decision you are informing. "Will the central bank cut at the next meeting?" is a topic, not a question. A researchable version fixes the actor, the action, the threshold and the deadline: which committee, which decision, measured by which published source, by which date. Writing this first stops you from quietly adopting whatever definition the first contract you found happens to use.

2. Find every contract that claims to answer it

Search across venues rather than settling on one. You are looking for candidates, not a single answer. For each candidate, record the venue, the native contract identifier and the exact title. Identifiers matter more than titles because titles get edited and reused, and because in three months you will want to know precisely which contract you were reading.

3. Read the contract, not the headline

This is the step most people skip. For each candidate, open the venue rules and note five things: what resolves it to yes, what resolves it to no, which source is authoritative, when trading closes, and what happens in an ambiguous or delayed case. Two contracts that disagree by eight cents often turn out to be asking slightly different questions once you read the settlement clause.

4. Record the state, with context

Capture the quote with the things that qualify it: the timestamp, the venue, the price convention, and some indication of activity such as recent volume or open interest. A price with thin activity behind it is a weaker signal than the same price with depth behind it, and a screenshot without a timestamp is close to useless a week later.

5. Decide what would change your mind

Before the next data release, write down the observation that would move your view and by roughly how much. This turns the market into an input rather than a verdict, and it gives you something to check yourself against later. Keep it in the same note as the contract identifiers.

A checklist you can reuse

Common mistakes

• Treating a price difference between venues as a free trade. It is a prompt to read the contracts, and it excludes fees, available size and settlement differences.
• Comparing a contract with a same-day deadline against one with a month-end deadline as though they measure the same thing.
• Reading a thinly traded contract as a confident consensus.
• Converting a price straight into a probability without noting the convention the venue uses and the costs that sit around it.
• Losing the identifier, so the note cannot be audited later.

Where Riddle fits

Steps two and four are the mechanical parts, and they are where a tool helps. Riddle covers Kalshi, Polymarket and Limitless, groups the contracts that appear to address the same question, and keeps the venue and identifier attached so you can go and read the source rules yourself. Grouping is a starting point for review, not a verified statement that two contracts are equivalent: step three is still yours.