How prediction markets work: prices, payouts and a paper walkthrough

Last Updated: 2026-10-26

A prediction-market position pays according to a written event rule. Knowing the headline is not enough: “a candidate wins”, “an outlet calls the race” and “the result is certified by a deadline” can settle differently. Start with that distinction before interpreting a price or connecting a wallet.

This guide combines our overlapping introductions. The examples are synthetic, not personal trades or a forecast of income. The Polymarket documentation was checked on the date above; account eligibility, supported funding routes and fees must be checked for the specific product and market.

Follow the money through one position

Assume an ordinary binary contract pays $1 per winning share and $0 per losing share, with no exceptional split payout. You buy 100 YES shares at $0.42, paying $42 before costs.

Scroll sideways to see every column. Keyboard: focus the table region and use the arrow keys.

The winning return is $58/$42, about 138.10%, not automatically 100%. The early sale returns about 85.71%. Neither sale price nor a buyer for the whole quantity is guaranteed. Deduct actual trading, funding and network costs separately. If entry and exit figures are actual fills, their difference already includes the spread you crossed; do not subtract that same spread twice.

In a standard complete set, one YES plus one NO has a combined $1 terminal payout. That is per matched pair, not $1 for an entire market. Polymarket distinguishes splitting collateral into a complete set, merging an equal pair, and redeeming resolved positions. These operations differ from buying through an order book. See its positions documentation. Check the applicable collateral and position system; this is not a deposit-address instruction.

A displayed price is not a promise or a fill

Suppose the best YES bid is $0.40 for 50 shares and the best ask is $0.42 for 30 shares. A midpoint of $0.41 does not let you buy 100 shares for $41. Additional asks may be more expensive; a limit order can remain partly or wholly unfilled. Daily trading volume does not establish current executable depth.

Likewise, YES and NO asking prices need not add to exactly $1. Buying both at $0.52 costs $1.04 for a standard $1 combined payout before costs. Redemption parity is different from two available purchase quotes. Do not combine contracts with different deadlines or resolution rules and call the gap guaranteed arbitrage.

Polymarket's trading overview separates signed orders, matching and settlement. Record the order ID, filled quantity and remaining open amount. A submission confirmation is not proof of a completed fill; a cancellation request needs a confirmed outcome too.

Use probability as an estimate with uncertainty

A 42-cent price is often read as an implied probability, but it is not a measured 42% chance. Liquidity, costs, access and differing beliefs affect prices. A conditional forecast also does not establish that one event causes another.

For the synthetic $1/$0 contract, if your estimated probability is 50% and the purchase price is $0.42, expected gross profit is $0.08 per share. If the plausible probability range is 35–55%, that figure ranges from −$0.07 to +$0.13 before costs. Uncertainty can change the decision; an exact-looking estimate does not remove it.

A five-cent share breaks even at one win in twenty identical trials before costs, not one in nineteen. This is payoff arithmetic, not a claim that such trials are independent or that the event really has a 5% chance.

Decide from today's value, not only the entry price

If you could now sell a share for $0.85, continuing to hold for a $1 payout risks that $0.85 current value for $0.15 further upside. Your original purchase price still matters for accounting; it does not make holding a risk-free trade. Compare your revised probability, executable exit price, remaining time and settlement uncertainty.

Separate markets can depend on the same outcome. Writing four different titles in a journal does not diversify that exposure. Four losses of 10% of the then-current bankroll leave 65.61% of the starting amount; four fixed losses of 10% of the initial bankroll leave 60%. Name the denominator instead of presenting either rule as protection from ruin.

Practice the full lifecycle without funding it

Choose one public market and complete this paper record:

1. Save its exact rule, deadline/time zone, resolution source and ambiguity provisions.
2. Record the time, bid/ask depth, planned quantity, fee assumptions and a price limit. Mark unavailable inputs as unknown.
3. Write a probability range and the evidence that would change it. Keep the original forecast when new information arrives.
4. Simulate fills only at available quotes and size. Track unfilled orders; do not count a touched price as a certain execution.
5. Record exit proceeds or the actual settlement outcome, all assumed costs and any disputed period. Separate realized results from open-position marks.

Polymarket's resolution documentation explains that the rule and resolution source govern the result. A news event, proposed answer, final payout and redeemed balance are different checkpoints. Some cases can have split payouts. Do not assume instant or automatic redemption in every interface.

Before any later funded use, verify eligibility and the precise product, official domain, supported asset/network, displayed receiving address, approval scope and withdrawal process. Never share a seed phrase. A wallet or collateralized contract does not eliminate issuer, software, oracle, access or key-loss risks. Do not increase a position merely to make fees look smaller.

Disclosure: Educational analysis, not financial, legal or tax advice. Positions can lose their entire cost. This site may earn commissions from identified affiliate links elsewhere; this guide provides no tested-return claim or paid-platform ranking.