Guide

How Prediction Markets Make Money

A prediction market exchange does not win when you lose. That is the structural difference between a venue like Kalshi or Polymarket and a sportsbook, and it explains nearly everything about how these platforms make money. An exchange matches one trader who thinks an outcome is likely against another who thinks it is not, and it takes a small, transparent cut of the trade — not a margin baked into the price. This page walks through where that money comes from: transaction fees and the maker-taker model, the bid-ask spread, what the exchange does and does not earn on deposits and settlement, and how Kalshi and Polymarket each monetize as of mid-2026. It also explains where the familiar "$1 per winning share" payout comes from, and why that number is fixed rather than a number the house chooses. Prices on these markets are crowd-implied probabilities that can be wrong, and nothing here is financial advice.

The exchange is a marketplace, not your counterparty

The first thing to understand is who you are trading against. On a traditional sportsbook, the book is the other side of your wager. It sets the odds, accepts your action, and profits from the "vig" (also called vigorish or juice) — a margin built into the price so the book comes out ahead regardless of which side wins. A typical point-spread line of -110 on both sides is the classic example: the extra cost above even money is the book's cut, and it works out to roughly a 4.5 to 5 percent edge on a standard line, and more on props and parlays. The sportsbook's revenue and the bettor's loss are two sides of the same coin.

A prediction market exchange works differently. It does not take the other side of your trade. When you buy a YES share, someone else on the platform is selling it (or buying the matching NO side), and the exchange simply matches the two orders and records the trade. Its revenue is a separate, disclosed transaction fee — closer to the commission a stock brokerage or a betting exchange charges than to a bookmaker's margin. Because the venue is neutral to the outcome, it has no incentive to shade prices against you. The price you see is set by the order book — by what other traders are willing to pay — not by a house trying to protect a margin.

This is why the money-weighted probabilities on these markets are useful as a news signal in the first place: the number reflects what participants with capital at stake collectively believe, rather than a line engineered to balance a book. That signal can still be wrong, thinly traded, or moved by a few large players, but the mechanism producing it is a marketplace, not a counterparty.

Trading fees and the maker-taker model

The core revenue line for a prediction market exchange is a per-trade transaction fee. The detail that surprises newcomers is that the fee is usually not a flat percentage — it scales with the price of the contract, and it is lowest exactly where you might expect a fee to be highest.

Kalshi's published taker fee, as of 2026, is $0.07 x C x (1 - C) per contract, where C is the contract price between $0.01 and $0.99. That C x (1 - C) term means the fee peaks when a market is a coin flip. At a 50-cent price the taker fee is about 1.75 cents per contract — its maximum — and it shrinks toward a fraction of a penny as the price moves toward 1 cent or 99 cents. In effect, the more lopsided (more certain) the market, the less you pay to trade it. Polymarket adopted a similar shape in 2026: its taker fee follows C x feeRate x p x (1 - p), with the rate set per category, and it likewise peaks at the 50 percent price.

The "maker-taker" split decides who pays. A taker is someone who removes liquidity by crossing the spread to hit an order already resting on the book; a taker pays the full fee. A maker is someone who posts a limit order that sits on the book and waits to be filled; the maker is adding liquidity, so the platform charges them far less — Kalshi's maker fee is 25 percent of the taker fee (about 0.44 cents at the 50-cent price) — or even pays them a rebate, as Polymarket does through its maker rebate program. The logic is the same on a stock exchange: makers make the market deeper and tighter, so they are subsidized; takers consume that depth, so they pay. None of this is advice on how to trade — it simply explains where the platform's fee comes from.

The bid-ask spread: a cost you pay to traders, not the house

Fees are not the only cost of trading, but the other main one — the bid-ask spread — generally does not go to the exchange. The spread is the gap between the highest price someone is currently willing to pay (the bid) and the lowest price someone is willing to sell at (the ask). If YES shares are bid at 47 cents and offered at 50 cents, a trader who wants in immediately buys at 50 and a trader who wants out immediately sells at 47; that 3-cent gap is the cost of demanding instant execution.

In a pure order-book exchange, that spread is captured by the market makers on the other side of those orders — the traders posting limit orders — not by the platform. This is a real distinction from a sportsbook, where the equivalent margin is the house's. On a deep, heavily traded market the spread can be a single cent or less; on a thin market it can be wide, which is itself useful information about how confident or liquid the crowd's view really is.

Some venues, particularly in their early form, used an automated market maker (an algorithm that always quotes a price from a pooled reserve) instead of, or alongside, a pure order book. There the spread and pricing behave differently, but the principle holds: the exchange's own take is the disclosed fee, while the spread is the price of liquidity set by whoever is on the other side. When you compare the "cost" of two platforms, the honest comparison is fee plus typical spread, not the headline fee alone.

Deposits, withdrawals, and where the $1 payout comes from

Beyond trading, exchanges can earn — or choose not to earn — on the money sitting on the platform. On Kalshi, ACH bank transfers in and out are free, while wire transfers carry a fee in the roughly $25 to $30 range (often charged by the bank rather than Kalshi itself), and there is no separate settlement fee for holding a contract to expiry. Polymarket charges nothing for USDC deposits or withdrawals itself, though the blockchain networks and third-party on-ramps it relies on can impose their own costs. A quieter revenue source is the interest, or net interest margin, an operator can earn on the pooled customer funds and collateral it holds — money parked on the platform that the operator can invest while it sits there. Increasingly, the data itself is a business: Polymarket has partnered with Intercontinental Exchange (ICE) to distribute real-time market-sentiment data to institutions, a revenue line with no per-trade fee attached.

The payout side is the part that confuses people most, and it is the simplest. Every contract is structured so that one full share is worth exactly $1.00 at resolution. If the event resolves YES, each YES share pays $1.00 and each NO share pays $0.00; if it resolves NO, the reverse. That $1 ceiling is not a number the house pays out of its own pocket — it is the total of what the two sides put in. A buyer who paid 60 cents for a YES share and a buyer who paid 40 cents for the matching NO share have together funded exactly $1.00; the winner collects that dollar, the loser collects nothing, and the exchange has already taken its small fee at the moment of the trade. Because every YES share and its matching NO share always sum to $1, the price of a YES share reads directly as an implied probability: 60 cents means the market collectively prices the outcome at roughly 60 percent. Those implied probabilities can be mispriced or move sharply on new information, and none of this is a recommendation to trade.

How Kalshi and Polymarket each monetize and settle

The two largest venues make money the same way in principle but differ in the details, partly because they are regulated and built differently.

Kalshi is a CFTC-regulated US exchange. Its revenue is the transaction fee described above, and its markets are cash-settled in US dollars: each contract names "source agencies" in its terms — the official data or governing body whose result determines the outcome — filed with the regulator as part of Kalshi's self-certification. When the underlying result is known, the exchange confirms it against that published rule and credits winning accounts, typically within hours. Clearing runs through Kalshi's registered clearing entity, and disputes are handled inside that regulated framework rather than by a vote.

Polymarket settles its core international markets on-chain in USDC and, as of 2026, charges category-based taker fees while rewarding makers — with some categories, such as geopolitics and world events, kept fee-free. It resolves outcomes through the UMA optimistic oracle: after a market closes, anyone can propose the result by posting a bond, a challenge window opens, and if no one disputes it the market settles; disputed cases escalate to token-holder voting and can take longer. Polymarket's US route runs through a CFTC-registered exchange it acquired to re-enter the American market. The practical takeaways: read each market's resolution rule before you trust the number, expect a delay between an event happening and a market paying out, and remember that resolution can occasionally be contested — all reasons a market price is a probability estimate, not a settled fact, and not financial advice.