Prediction Market Arbitrage: Polymarket vs Kalshi

What is cross-exchange arbitrage between Polymarket and Kalshi?

Cross-exchange arbitrage is the practice of pricing the same real-world question on two venues at once and locking the difference before it closes. On a prediction market, a share pays $1 if the outcome happens and $0 if it does not, so the price of a YES share is just the market's implied probability in cents. When Polymarket lists a contract at 58 cents and Kalshi lists the mirror of that same event at 46 cents, the two order books disagree about the same future. That disagreement is the raw material of the strategy.

The core move is simple to state. You buy YES on the venue where it is cheaper and buy NO on the venue where NO is cheaper, sizing both legs to the same number of shares. If the combined cost of one YES share plus one NO share comes in under $1, and both contracts resolve to the identical real-world result, then exactly one leg pays $1 at settlement and you keep the spread. The position is directionally flat. You are not trading a view on the outcome, you are trading the gap between two prices for the same outcome.

This is a copy trading discipline built on intelligence, not blind copying. The goal is not to chase whatever a single account did, it is to read where informed order flow has already repriced one venue and route your own legs to the side that has not caught up yet. Treat every apparent spread as a claim you have to verify, because the number on the screen is only real if the two contracts truly resolve on the same criteria.

Why does the Polymarket vs Kalshi price gap exist?

Two exchanges, two separate order books, two different crowds. Polymarket settles in USDC and draws a crypto-native base. Kalshi is a US regulated venue that settles in dollars and draws a different set of participants. The same event can carry a different consensus on each because the people pricing it are not the same people, and money does not move instantly between the two. That structural separation is why the spread is not arbitraged away the moment it appears.

Edges live in the lag. When news breaks, smart money tends to hit one venue first, usually the deeper or faster one for that category, and the other venue trails by seconds or minutes. During that window the two prices describe the same future at two different probabilities. Sharp traders who watch both books at once see the divergence open before the slower side reprices. On Kalshi, that informed activity shows up as anonymous flow, shifts in size and price that you read as a signal, never as a named person.

Fee schedules, gas and withdrawal frictions, and uneven liquidity all widen the band in which a gap can persist. Because moving capital across venues is not free or instant, the market tolerates a standing spread rather than collapsing it to zero. Your job is to find the moments when that spread is wide enough to clear the real costs, and to act while it is still open.

How do you lock a cross-exchange spread step by step?

If you are learning how to trade prediction markets across venues, start on the Divergence & Arbitrage board. It scans matched contracts across Polymarket and Kalshi and surfaces pairs where YES on one venue plus NO on the other totals under $1. The board routes each leg to the best available price so you can see the combined cost and the raw spread before you commit a dollar. Sort by the widest gaps, then discard anything that does not survive the checks below.

Verify the resolution criteria before anything else. Two contracts can share a headline and still settle differently on the edge cases: the cutoff time, the data source, what counts as the event, how ties or cancellations are handled. If the wording diverges, the spread is not an arbitrage, it is two different questions that happen to look alike. This single check is what separates a locked spread from a hidden directional position.

Read the flow that opened the gap. Use the Sharp Score to gauge how informed the activity behind a move is, and open the Live Feed to watch order flow arrive in real time. Cross-check against Consensus to see where the broader book has settled, and use the Master Wallet view to understand how the most consistently sharp order flow is leaning. On Kalshi this is always anonymous flow read as intelligence, not a person you are copying.

Pre-position, then execute both legs close together. Because withdrawals are slow, you want capital already sitting on both venues so you can fill YES here and NO there in the same short window. Fill the less liquid leg first, since that is the one more likely to move against you. Set Alerts so the board pings you when a pair crosses your minimum net spread, and add the contracts you are stalking to your Watchlist and Tails so you are not manually refreshing two exchanges.

For adjacent setups, the same read carries over to sibling boards. The Fade Board flags where over-extended flow is likely to revert, the Insider Radar surfaces unusually informed early activity, and the Weather Edge covers climate-settled contracts where the resolution source matters even more than usual. Each one is a different lens on the same underlying question of where price and information have separated.

How much capital should you split across two venues?

As a prediction market strategy, cross-exchange arbitrage needs capital parked on both sides at once, so sizing starts with the split, not the trade. Decide a total amount you are comfortable committing, then divide it so each venue can carry its share of both YES and NO legs without you needing to move funds mid-trade. Keep a reserve on each side for the pair you did not see coming. Never commit money you cannot afford to lose, and never let the strategy pull in rent or bills money.

Size each spread by the net edge after all costs, not the gross gap on the screen. A pair showing a 4 cent raw spread that costs 3 cents in fees, gas, and slippage is a fraction of a cent of real edge, and that is before a leg slips on you. Cap any single pair at a small fraction of the venue balance so one bad fill or one resolution surprise cannot dent the book. Many small, well-verified spreads beat a few large ones.

Treat locked capital as a real cost. Both legs stay tied up until the event resolves, which can be days or weeks, so the same dollar cannot chase the next gap while it is committed. Factor that opportunity cost into which spreads are worth taking. This is responsible trading, sizing to survive a long string of ordinary outcomes rather than to maximize any one of them.

Is prediction market arbitrage actually profitable?

The honest answer is that the edge is real but thin, and costs decide whether it survives. Consider a hypothetical pair: YES on Polymarket at 47 cents and NO on Kalshi at 49 cents, a combined 96 cents for a position that pays $1 at resolution. On paper that is 4 cents of gross spread. Subtract trading fees on both legs, gas or transfer friction, and any slippage on the fill, and the net can easily fall to a cent or less. The math only clears at scale and with disciplined execution. This example is hypothetical and is not a prediction of any specific result.

Profitability also depends on how often clean, verified pairs actually appear. Wide spreads that pass the resolution-criteria check are not constant, and the widest ones tend to be the least liquid, which caps how much you can deploy. Any figure you see about historical spreads is context, not a promise. Past performance does not guarantee future results, and no version of this strategy is a guaranteed profit.

The realistic framing is a grind of many small, verified spreads where the discipline is in what you decline. The traders who make this work are not the ones taking every gap, they are the ones who reject the pairs that fail a resolution check, refuse the ones where net edge does not clear costs, and only fire when both legs can be locked close together.

What are the failure modes and resolution traps to watch?

The most expensive failure is a resolution mismatch. If the Polymarket and Kalshi contracts settle on different cutoffs, different data sources, or different definitions of the event, you were never flat. You were holding a directional position that only looked hedged, and one leg can pay while the other does not. This is why the resolution-criteria check is not optional. It is the strategy.

Leg risk is the next one. If you fill YES on one venue and the NO leg moves before you complete it, you are left half-hedged and exposed to the outcome. Fill the thinner leg first, keep both windows tight, and be willing to unwind a single filled leg rather than chase a spread that has already closed. Alerts and the Watchlist exist so you are ready to act in that short window rather than reacting late.

Then there are the frictions that quietly erode the edge: capital locked until resolution, slippage in thin books, withdrawal delays that strand funds on the wrong venue, and platform or counterparty risk on either side. None of these are reasons to avoid the strategy, they are reasons to size for them. Trade responsibly, keep records, and never stake money you cannot afford to lose. The spread is only worth taking when the edge survives every one of these costs with room to spare.