A trader scrolls past forty markets on Kalshi and stops on the forty-first. Not because it's the biggest pot, or the one in the headlines, but because the yes side is sitting at 71 cents and a near-identical contract on Polymarket is at 64. That seven-cent gap is the moment edge-hunting begins. The question is whether it's real, and whether you can collect it before someone faster does.

Mispricing in prediction markets is rarely the kind of thing a casual scroll surfaces. The obvious gaps close in minutes. The persistent ones live in the corners: thin order books, awkward resolution criteria, contracts that traders simply forgot were live. Knowing where to look is a skill, and like most trading skills it is mostly about ignoring the noisy 95% of the screen to find the quiet 5% that actually pays.

The cleanest edge is the gap nobody is watching

The biggest contracts on any platform are also the most efficient. The 2028 US presidential market on Polymarket has tens of millions of dollars in volume and dozens of professional traders monitoring every tick. The probability of you spotting a mispricing there before a hedge fund algorithm does is roughly nil.

The edge lives further down the page. A municipal election in a mid-sized US state. A central bank decision in a country most traders can't place on a map. A reality TV finale that resolves on a date three weeks out. These are markets where the liquidity is thin enough that a 50-cent contract can sit at 50 cents for eleven hours while the underlying reality shifts noticeably, because nobody bothered to update the order book.

That is the appeal. It is also the trap. Thin markets are mispriced precisely because they are hard to trade out of, and a contract you bought for 50 cents that you can't sell back is not a contract you priced correctly, it is a contract you are now stuck with until resolution.

Where structural mispricing actually comes from

Genuine mispricing falls into a small number of recognisable shapes. The first is the cross-platform spread, where the same underlying outcome trades at meaningfully different prices on Polymarket, Kalshi, Betfair Exchange and Smarkets. Sometimes this is a real edge. More often it reflects different resolution wording, different fee structures, or different access regimes (a market only US traders can enter will price differently to one only EU traders can). Our guide to Polymarket arbitrage walks through which gaps are tradeable and which are illusions.

The second shape is the stale book. A market opens, traders pile in around an opening narrative, then the news cycle moves on and the contract sits frozen at last week's consensus. New information arrives. The order book doesn't. The mispricing here is not against another market; it is against reality.

The third shape is resolution ambiguity. A market asks whether a politician will be "in office" on a given date, but the wording does not specify acting versus confirmed, interim versus elected. Half the traders are betting on one interpretation, half on the other, and the price reflects an average that matches neither. Smart traders read the resolution criteria carefully, decide which interpretation the platform will actually use, and take the side everyone else has mispriced.

The fourth, and the one most retail traders miss, is the favourite-longshot bias. Across decades of betting data, longshots are systematically overpriced and heavy favourites systematically underpriced. A contract trading at 4 cents will resolve YES less than 4% of the time on average; a contract at 96 cents will resolve YES more than 96%. The gap is small, often a couple of cents, but it compounds across hundreds of trades.

Reading the order book like a tell

The price you see on a market summary screen is a lie. Or rather, it is one number describing a structure that has more interesting numbers buried inside it. The order book is where the real story lives.

A market trading at 60 cents with $200,000 of bids stacked between 58 and 60, and only $4,000 of offers between 60 and 62, is not a market that wants to stay at 60. It is a market with a roof someone forgot to reinforce. Two well-sized buys and it moves to 65, no news required.

The inverse also holds. A contract sitting at 30 cents with a thick wall of sell orders at 31 and 32 is being capped by someone (often a market-making bot) who has decided the fair price is no higher than that. If the underlying reality justifies 40 cents, you are looking at a structural mispricing being maintained by a single counterparty, and the question becomes whether they will defend the cap or fold when the orders keep coming.

This is why understanding liquidity in prediction markets is more useful than memorising any single trading strategy. The order book tells you not just what the price is, but how much it would cost to move it, and that is the actual information you need.

Polls, news cycles and the lag that pays

Professional traders talk about "information latency" and they mean it literally. A poll drops at 6am. The market reacts by 6.05am for high-profile contracts and somewhere between 8am and never for lower-profile ones. The window between the information existing and the market reflecting it is where retail edge lives.

The trick is knowing which information sources the market reads and which it doesn't. Polymarket traders follow Nate Silver and the major US polling aggregators. They are less reliable about regional polls in non-English-language jurisdictions, council-level UK by-elections, and obscure regulatory filings buried in 200-page documents. If you have a habit of reading something the bots don't, that habit is worth money.

This is also where the polls-versus-markets comparison gets practical. Polls and markets diverge most when one of them has stale information. Spotting which one is stale, and how long the divergence has been open, is most of the edge.

The mispricings that aren't

Reasonable people argue that any gap between two prices is an edge. The regulation, mechanics and order flow disagree. A 10-cent spread between Polymarket and Betfair Exchange on the same election might look like free money, but the actual trade involves USDC on one side, GBP on the other, a 30 minute fiat transfer, settlement risk on both venues, and a fee drag that eats half the headline gap before you've even opened a position.

Worth flagging: most published "prediction market arbitrage" content does not net out fees, settlement time, withdrawal costs, or the simple risk that one venue resolves the market differently to the other. Two contracts asking what looks like the same question can resolve to opposite outcomes if the small print differs by a single clause.

The other false-positive is the longshot that "should obviously be higher". A contract at 2 cents on an unlikely outcome looks cheap, but if you buy 5,000 contracts at 2 cents and they all settle no, you are out the full $100, and the expected value math only works if your read on the true probability is sharply better than the market's. Most retail traders overestimate how often this is true.

A workflow that finds real gaps

The edge-finding process that actually works is unglamorous. Pick a vertical you know well (a specific sport, a specific political jurisdiction, a specific scientific field). Open three or four platforms side by side. Scan for contracts where the price has not moved in the last 24 hours despite material news. Cross-check the resolution criteria carefully, read the order book to understand what it would cost to move the price, and only then size a position you can afford to be wrong about.

Most trades you screen will fail one of those tests. The resolution wording will be ambiguous in a way that kills the edge. The order book will be too thin to enter without moving the price against yourself. The news will turn out to already be priced in by a 3am move you missed. That is normal. Edge-finding is mostly rejection; the actual trades are rare and almost always smaller than feels exciting.

If this is genuinely how you want to spend your trading time, our overview of how to make money on prediction markets is worth reading alongside this one. It covers the broader question of which traders actually finish the year ahead, and it is more sobering than most strategy guides admit.

Where iPredicta fits

iPredicta is a discovery platform for prediction markets, aggregating contracts across Polymarket, Kalshi, Betfair Exchange, Smarkets and the regulated UK venues so traders can scan multiple order books in one place rather than tabbing between platforms. The point is not to tell you which contract to trade. The point is to surface the contracts you wouldn't otherwise see, so the edge-finding workflow above takes ten minutes instead of two hours.

Frequently asked questions

What is a mispriced prediction market?

A mispriced prediction market is one where the contract price diverges meaningfully from the true probability of the underlying event, in a way you can identify and trade. The divergence can come from stale order books that haven't updated to new information, ambiguous resolution wording that splits traders between two interpretations, or simple lack of attention on lower-profile contracts. Genuine mispricing is rare on the biggest markets because professional traders monitor them constantly. It is more common on thin, obscure or recently opened contracts. Spotting one is the easy part; getting a meaningful position in and out at the prices you want, given the thin liquidity that usually causes the mispricing in the first place, is harder.

Are cross-platform spreads always a trading opportunity?

No, and assuming they are is the most common beginner mistake. A 10-cent gap between the same contract on Polymarket and Betfair Exchange often disappears once you account for fees, currency conversion, settlement time, withdrawal costs and the risk that the two venues resolve the question differently. The resolution wording is the critical check. Two markets that look identical can settle to opposite outcomes if one defines "in office" as elected and the other as acting, or if one counts a recount while the other doesn't. Read the small print on both venues before assuming the spread is real money. The arbitrage primer covers this in more detail.

How thin is too thin when looking for mispriced markets?

If you cannot exit your position at roughly the price you entered it, the market is too thin to trade regardless of how obviously mispriced it looks. A useful rule is to check the order book depth at the price you're considering, then check it again 5 cents away in both directions. If a sensible position size would move the price by more than a couple of cents against you on the way in, the round-trip cost will eat most of your theoretical edge. Thin markets are also more likely to have wide bid-ask spreads, low daily volume, and long delays between trades, all of which compound the practical problem of getting your money back out at the price the model says it's worth.

Do retail traders really find edges that professionals miss?

Yes, but only in specific corners. Professional traders and market-making bots dominate high-volume English-language political and macroeconomic markets, where retail edge is essentially zero. Retail traders find genuine edges in markets that require specialist knowledge the bots don't have: a specific sport, a non-English-language political jurisdiction, a niche scientific or technical question, an obscure local election. The edge is not "I'm smarter than the algorithms". It is "I read a source the algorithms don't read, and I noticed something before the price did". That is a defensible, repeatable edge if you stay in your specialism and resist the temptation to trade the headline-grabbing markets where you have no information advantage.

How long do mispriced markets typically stay mispriced?

It depends entirely on visibility. A mispricing on a flagship Polymarket contract closes in minutes, sometimes seconds, because professional traders and arbitrage bots are watching constantly. A mispricing on a low-volume contract in a niche category can persist for hours or even days, because nobody is looking. The trade-off is that the same thinness that lets the mispricing persist also makes it hard to enter and exit cleanly. Long-lived mispricings on thin markets are usually long-lived precisely because they are practically untradeable at scale. The sweet spot is a mispricing on a medium-liquidity contract that you spot before the broader market does, and where you can size a position big enough to matter without moving the price.