Open the same real-world event on two prediction market platforms and you will often see two different numbers looking back at you. Same headline, same outcome, same rough deadline — and yet one venue prices it at, say, the low 60s and another prices it in the low 70s.
If you already understand how prediction market probabilities work — that a price is a market-implied probability, not a guarantee — the natural next question is what to do when two venues disagree on that probability for what looks like the same question. That gap is prediction market divergence, and it shows up constantly once you start looking: a polymarket vs kalshi odds mismatch on the same political outcome, a few points of prediction market disagreement on a Fed decision, or a much wider polymarket kalshi price difference on a niche market that one venue's crowd has barely touched.
This guide is about reading that gap correctly. Same event different odds situations are one of the most useful cross platform prediction markets signals available — and one of the most commonly misread. Most of the time a prediction market spread is telling you something structural (fees, access, wording, liquidity), not handing you free money.
TL;DR
- The same event is often listed as separate contracts on separate platforms, each with its own crowd, fees, rules, and liquidity — so
market implied probability differencesare normal, not a bug. - Five structural causes explain most gaps: fees, user-base/access restrictions, resolution wording, collateral/currency, and liquidity depth. Resolution wording is the one most likely to bite you.
- Divergence is signal when it tracks real news and both sides are liquid enough to trust. It is noise when one side is thin, stale, or quietly pricing a different question.
- "Same event" is very often not "same contract." Read the resolution rules on both platforms before treating a gap as meaningful.
- Textbook cross-venue
prediction market arbitragerarely survives fees, withdrawal friction, split capital, and resolution-wording basis risk once you try to actually execute it. - Treat divergence as a research prompt, not a profit formula. CoinRithm's Compare view and Today view exist to help you do that research in one place.
- This is educational content, not financial or trading advice.
Why the Same Event Prices Differently Across Platforms
A prediction market price is the implied probability of one specific crowd, trading one specific contract, under one specific set of rules. When the "same" real-world event is listed on Polymarket, Kalshi, Limitless, and others at the same time, each listing is its own separate market — its own order book, its own traders, its own fine print. Market implied probability differences between venues are the expected output of that setup, not evidence that something is broken. Five structural factors explain most of what you will see.
Fee Structures
Taker fees, maker rebates, fees charged only on net winnings, and withdrawal or network costs all vary by platform. A venue with higher effective fees needs its quoted price to sit slightly further from "fair" before a trader will bother, because part of what looks like edge is actually fee drag waiting to be paid. Two platforms can show a real gap in raw probability while showing almost no gap once fees are netted out on both sides — or the opposite, where a small quoted gap disappears entirely after costs.
User-Base and Access Restrictions
Each platform's price is set by whoever is actually allowed to trade there. A regulated, KYC'd, largely US-based user base and a global, crypto-native, largely non-US user base are not the same crowd, and they don't always have the same information, risk appetite, or reaction speed to the same news. Access restrictions are not just a compliance detail — they change who is setting the price in the first place, which is a real reason for prediction market disagreement that has nothing to do with either venue being "wrong."
Resolution-Wording Differences (the Big One)
This is the reason experienced traders check first and beginners check last. Two markets can share almost the same title and still not be the same bet: one may resolve against a specific government data release, another against a news wire report of that same data; one may cut off at 11:59pm in one timezone, another at midnight in a different one; one may have explicit rules for a cancelled or postponed event, another may leave that ambiguous.
Same event different odds very often means "same event, different contract." A gap that looks like a mispricing can simply be two markets correctly pricing two subtly different questions. This is exactly the territory covered in How Prediction Markets Resolve — read it before you treat any cross-venue gap as an opportunity rather than an artifact.
Collateral and Currency Differences
Some venues settle in on-chain stablecoins, some in fiat through a regulated account, and some in play-money points on a forecasting platform rather than a real-money exchange at all. Collateral choice quietly changes the price: stablecoin depeg risk, smart-contract risk, and on/off-ramp friction are all real costs that a purely probability-based comparison ignores. A price expressed in USDC on one venue and USD on a regulated venue are not perfectly interchangeable numbers even when the percentage looks identical.
Liquidity Depth and Market-Maker Presence
A market with real depth and active market-making absorbs single trades without much price movement, so its quote is closer to a genuine consensus. A thin market can be moved by one order, which means its "probability" may reflect one trader's opinion rather than a crowd's. Comparing a deep, actively made market against a thin one and calling the gap "divergence" is often comparing a signal to noise.
When Divergence Is Signal vs When It's Noise
Not all gaps deserve the same reaction. The useful distinction is whether the disagreement is telling you something real about the event, or something mundane about the market's plumbing.
Divergence tends to be signal when:
- one venue's crowd reacted to breaking news first, and the gap is closing over time as the slower venue catches up
- both sides are reasonably liquid, so neither price is a single-trade accident
- the gap is widening on rising volume on both venues, not just one quiet order book drifting
Divergence tends to be noise when:
- one side is thin — a handful of trades can move a quote without reflecting any real change in belief
- one side is stale — no recent trade, so the "price" is really yesterday's opinion, not today's
- the two markets are subtly different questions once you read the resolution rules closely
As an illustration only: suppose a debt-ceiling deadline question is priced at 62% on one venue and 74% on another. That gap could mean one venue's crowd is genuinely ahead of the news. It could just as easily mean the 74% market is illiquid and hasn't traded in hours, or that its resolution clause defines the deadline a few days later than the other venue's does. The number alone can't tell you which — you have to check.
The most important habit here is refusing to assume "same event" means "same contract." Verify it every time, on both source platforms, before drawing a conclusion from the gap.
The Arbitrage Honesty Section
A wide, clean gap between two matched markets naturally suggests a version of prediction market arbitrage: buy the cheaper side of the outcome on one venue, buy the opposite side on the other, and lock in a spread regardless of the result. On paper, this looks like free money. In practice, it rarely is — for retail traders, for several concrete reasons:
- Fees eat the theoretical edge. Taker fees on both legs, and fees charged on net winnings on the winning side, can quietly consume most or all of a thin quoted gap.
- Withdrawal friction and funding lag. Moving capital onto and off of two different platforms — one of them possibly on-chain, one of them possibly a regulated fiat account — takes time and sometimes costs money in its own right.
- Capital sits in two places at once. A "hedged" position still requires posting collateral on both venues simultaneously until resolution, which is real capital tied up and real counterparty exposure to two platforms instead of one.
- Resolution-wording basis risk. This is the one that turns arbitrage into a loss. If the two contracts are not actually identical — different source of truth, different cutoff, different edge-case handling — then what looked like a hedge can resolve as two losing legs instead of one guaranteed spread. The "arbitrage" was never real; it was two different bets that happened to look alike.
- Access and legality per venue. Not everyone can legally hold both legs. A regulated, KYC'd venue and a globally accessible but US-restricted venue don't have the same eligible user on both sides — the trader who can theoretically capture the spread on paper may not be a trader who can actually open both accounts.
None of this means divergence is worthless — it means the honest use of a probability gap is as a research signal: it tells you where the crowd disagrees and prompts you to find out why. It is not a mechanical formula for guaranteed profit, and nothing in this article should be read as suggesting otherwise. This is educational content, not trading advice.
How to Read CoinRithm's Compare View and Today Page
CoinRithm's Compare view exists specifically to make cross-platform reading like this practical instead of a seven-tab exercise. Instead of listing markets separately, it matches the same underlying question across sources and groups the matched contracts into one cluster.
Each matched cluster shows:
- A divergence badge stating the size of the current gap in points, so the widest disagreements are easy to find without doing the arithmetic yourself.
- Two tabs — "All matches," ordered by match confidence, and "Divergence," which applies a minimum-gap threshold and ranks only the clusters where sources disagree the most.
- Decision-support signals on each card: match quality, spread reading (tight, moderate, or wide), a liquidity reading (high, medium, or low), and flags for a thin market or a market nearing resolution — the same structural factors this article walks through, surfaced directly on the card instead of left for you to infer.
The practical workflow: open the Compare view's Divergence tab first. For any cluster with a wide gap, check the liquidity signal before anything else — a thin-market flag on either side is usually the whole explanation. If both sides look liquid, go read the resolution rules on each source platform to confirm the contracts genuinely match. Only once a gap survives both checks is it worth treating as a real disagreement rather than an artifact.
The Today view is the discovery layer that feeds this workflow. It rolls the day up across every source at once — biggest moves, newly opened markets, markets expiring soon, markets that just resolved, and where volume is concentrated. Scanning Today first, then jumping into Compare for anything that looks interesting, is a faster way to catch a developing disagreement than checking each platform on its own schedule.
A Reading Checklist Before You Trust a Spread
Before treating any cross-venue gap as meaningful, run it through this checklist:
- Is the gap wide enough to matter after fees, or does it disappear once realistic taker fees and withdrawal costs are netted out?
- Is either side thin or stale? Check the liquidity signal before anything else — a single-trade price is not a consensus.
- Do the resolution rules actually match — same source of truth, same cutoff, same handling of edge cases — on both platforms, not just the title?
- Is the gap widening or shrinking, and does it track a specific, verifiable piece of news rather than random noise?
- Could you legally and practically access both venues, or is one of them off-limits to you — which makes the "arbitrage" framing moot from the start?
- If you're only trading one side, does that venue's price look mispriced against your own independent read of the event — not simply against the other venue's number?
If a spread survives all six questions, you've found a genuine disagreement worth researching further. If it fails any one of them, you've avoided treating an artifact as an opportunity.
How CoinRithm Fits In
CoinRithm does not execute real-money trades or hold custody of funds — it is a research and aggregation layer that sits on top of the underlying venues, including the two most-compared platforms, Kalshi and Polymarket.
Use the Compare view to find matched clusters and see the size of the gap in points before you form a view. Use the Today view to catch a developing disagreement while it's still moving rather than after the fact. For the full cross-platform research workflow — including how to weigh fees and liquidity alongside probability — read How to Compare Prediction Markets. Before treating any matched pair as the same bet, confirm how the markets actually resolve on each source platform.
If you want to pressure-test a read on a divergent market without risking real money, CoinRithm's paper-trading simulator lets you size a mock position on either side and see how it plays out before you ever commit real capital on the underlying platform.
Frequently Asked Questions
Why do Polymarket and Kalshi show different odds for the same event?
Because they are different contracts, not the same market split across two windows. Each has its own crowd, fee structure, access restrictions, collateral type, and — most importantly — its own exact resolution wording. Any one of those can produce a polymarket kalshi price difference even when both platforms are functioning normally.
Does a probability gap between platforms mean there's an arbitrage opportunity?
Not automatically, and often not at all. A wide gap is more commonly explained by a thin or stale market on one side, or by the two contracts quietly resolving on different criteria, than by a real, executable arbitrage. Treat a gap as a prompt to investigate, not as a signal to trade immediately.
What is resolution-wording basis risk?
It's the risk that two markets which look like the same question actually settle on different rules — a different source of truth, a different cutoff time, or different handling of edge cases. If you build a "hedge" across two contracts that aren't truly identical, you can end up with two losing legs instead of a locked-in spread. Always confirm resolution rules on both source platforms before assuming two markets are equivalent.
How wide does a spread need to be before it's worth investigating?
There's no fixed number that applies across every event, since it depends on liquidity, fees, and how confidently the two markets are matched. The useful test isn't a specific point threshold — it's whether the gap survives the reading checklist: real after fees, not explained by thinness or staleness, and backed by matching resolution rules.
Can I actually execute cross-platform prediction market arbitrage as a retail trader?
Rarely in the clean, guaranteed-profit form it's often described in. Fees on both legs, withdrawal and funding friction, capital tied up simultaneously on two platforms, and resolution-wording basis risk all cut into or eliminate the theoretical edge. Access and legality differences between venues can also mean you're not eligible to hold both legs in the first place. This isn't financial advice, and no spread should be treated as a reliable source of profit.
How does CoinRithm show probability divergence?
The Compare view matches the same question across sources into a cluster and shows a divergence badge with the size of the gap in points, plus decision-support signals for match quality, spread, and liquidity. Its Divergence tab filters to only the clusters where sources disagree the most, so the biggest gaps are easy to find and easy to sanity-check before you trust them.
Continue reading: Prediction Market Data API
Disclaimer: This article is for educational and informational purposes only. It is not financial, legal, or investment advice, and nothing here should be read as a claim that cross-platform arbitrage is reliable or guaranteed. Prediction markets involve financial risk — you can lose your entire position, on one leg or both. CoinRithm aggregates data for research and does not execute real-money trades. Fees, resolution rules, liquidity, and platform availability can change at any time; always verify current terms directly with each platform before trading.
