Question 1

How does cross-platform prediction-market arbitrage actually work, and why is the combined payout always exactly $1?

Accepted Answer

A binary event contract pays exactly $1 if the outcome happens and $0 if it doesn't, on every platform that lists it. If Polymarket and Kalshi both list the same real-world event — say, the same election, the same rate decision, the same game — then buying YES on one platform and NO on the other locks in one contract that resolves to $1 and one that resolves to $0, no matter which way the event actually breaks. You never have to guess the outcome, because you own both sides of it across two separate order books. The only variable that matters is what you paid to assemble that guaranteed dollar. If YES on platform A costs 42 cents and NO on platform B costs 45 cents, you've spent 87 cents to lock in a payout of $1, which is a 13-cent raw edge per contract before fees — thousands of these pairs and the edge scales linearly, which is exactly what this calculator is pricing.

Question 2

Why does this calculator assume no bid/ask spread, and what could go wrong with that in practice?

Accepted Answer

This tool takes each platform's YES price at face value and derives the NO price on platform B as 1 minus that YES price, which is only true if there's zero spread between the best bid and best ask and if your order size doesn't move the book. In real order books that's rarely exact — the price you actually get filled at, especially for the size needed to make small per-contract edges add up to meaningful dollars, is usually a few cents worse than the quoted mid, and thin markets can eat the entire edge in slippage alone. Beyond execution price, the two biggest practical risks are timing and wording. Timing risk is the gap between when you fill leg one and leg two — prices move in between, and a fast-moving event can close the edge or flip it negative before your second leg fills. Wording risk is worse and harder to see coming: Polymarket and Kalshi write their own resolution criteria for what counts as the event happening, and if the two contracts don't resolve on identical criteria — different deadlines, different data sources, different tie-breaking rules — you can end up with a contract that pays $0 on both legs or, in the ugly case, a mismatch where neither side pays what you expected. Always read both platforms' resolution rules side by side before sizing a real position, not just the headline market question.

Question 3

How does Kalshi's per-contract fee formula work, and why does it scale with price times (1 minus price)?

Accepted Answer

Kalshi's standard trading fee is roughly fee = feeMultiplier × contracts × price × (1 − price), where price is the contract's price as a fraction and the multiplier is about 0.07 on most markets. That price × (1−price) term is a parabola that peaks at exactly 50 cents and shrinks toward zero as the price approaches either 1 cent or 99 cents. The intuition is that Kalshi is pricing the fee off the notional risk being transferred, and a coin-flip contract at 50 cents has the most uncertainty — and the most capital genuinely changing hands relative to the eventual $1 payout — while a 5-cent or 95-cent contract is a near-settled bet where very little is actually being wagered on the outcome. For an arbitrage pair, this matters because the fee is charged independently on each leg based on that leg's own price, not on your combined edge, so a pair priced near 50/50 on both platforms pays close to the maximum fee bite on both legs simultaneously, while a highly lopsided pair (say 8 cents and 90 cents) barely gets touched by fees at all — which is exactly why some real-world arb opportunities that look attractive on raw price alone evaporate once Kalshi's fee curve is applied to the leg trading near the middle.

Question 4

How is this different from the single-platform prediction market calculator?

Accepted Answer

The prediction market calculator prices one position on one platform: you pay a contract price, it tells you your payout if you're right, your fee-adjusted breakeven probability, and the equivalent betting odds. That tool is about whether your own view of the probability is better than the market's — it's directional, and you can lose the whole stake if you're wrong. This calculator prices something structurally different: two offsetting positions on two different platforms for the exact same event, sized so that one side always pays $1 and the other always pays $0, regardless of which way the outcome goes. There's no view on probability involved at all — the profit or loss is locked in the moment both legs fill, and it comes purely from the gap between what you paid across both legs and the $1 you're guaranteed to collect. If you're only trading one platform, use the single-platform tool; if you've spotted the same event mispriced across two books and want to know whether hedging across them is actually profitable after fees, this is the one that answers that.

Calculateur d’arbitrage de marché prédictif

Dans quelle mesure l'arbitrage est-il sensible au prix de la plateforme B ?

Une couverture, un dollar garanti