The price is a probability — until you try to take it.
A prediction market turns a question into a share that pays €1 if the answer is yes and €0 if it is no. Because the payout is fixed, the price — always between 0 and 1 — reads directly as the market's estimate of the probability. That sentence is on every explainer. This page covers the part that isn't: where the number on the screen comes from, what it costs to trade against it, and the three places the shorthand breaks.
One share, all the way through
Take a market on a clear yes/no question. The Yes share trades at 62¢. You buy 100 of them for €62. If the answer turns out to be yes, they pay €100. If no, they pay nothing.
The conversion to the decimal odds a bookmaker would quote is exact: odds = 1 / price. And because Yes and No must sum to €1, buying No at 38¢ is the same trade as selling Yes at 62¢ — one position, two ways to describe it.
| Price | Decimal odds | Implied chance | Cost of €100 of shares |
|---|---|---|---|
| 10¢ | 10.00 | 10% | €10.00 |
| 25¢ | 4.00 | 25% | €25.00 |
| 50¢ | 2.00 | 50% | €50.00 |
| 62¢ | 1.61 | 62% | €62.00 |
| 80¢ | 1.25 | 80% | €80.00 |
| 95¢ | 1.05 | 95% | €95.00 |
The other thing a share gives you that a bookmaker's slip doesn't: you can sell it before the event resolves. If your 62¢ share trades at 80¢ because the news broke your way, you can take the 18¢ and leave. Whether anyone is there to buy it from you is the subject of the next section.
The price on the screen is the top of a ladder
There is no single "price". There is an order book: a list of buy offers (bids) and sell offers (asks), each with a size. The best ask is the cheapest anyone will currently sell to you; the best bid is the most anyone will currently pay you. Between them is the spread, where no trade happens. Most interfaces show you one number, usually the mid or the last trade — neither of which is a price you can actually get.
This is the single most useful thing to understand about these venues. Books on prediction markets are thin: the best price often holds only tens of shares. A market order takes whatever it needs, level by level, and the average price you pay is a weighted average of everything it ate — never the number you clicked.
Two consequences worth acting on:
- Size against depth, not against the price. Before placing a market order, look at how many shares sit at and near the touch. If your order is larger than the top two levels, you are the one moving the price.
- A market order needs a cap. Any tool worth using limits how far your order may walk up the book and cancels the rest rather than filling at any price. If yours doesn't tell you what that cap is, assume there isn't one.
The alternative is a limit order: you name a price and wait. You don't pay the spread — you might even earn it — but you may not fill at all, and in a market that is about to resolve, not filling is its own kind of loss.
What it costs
The spread
On the book above, you buy at 63¢ and could immediately sell at 61¢. That 2¢ round trip is 3.2% of a 63¢ share, paid before you are right about anything. On markets priced near the extremes the spread in cents may look tiny and be enormous in percentage terms: a 1¢ spread on a 4¢ share is 25%.
Fees
Polymarket sets a taker fee per market, and most markets are set to zero. Where one applies, two details matter. It is charged on top of what you spend rather than taken out of your winnings, so it eats into your budget before the order is sized. And it is scaled by min(price, 1 − price) — it bites hardest at even money and shrinks towards the extremes. A 10% market rate at a price of 31¢ costs 0.10 × 0.31 = 3.1% of the money you spend; the same rate at 5¢ costs 0.5%.
Minimums
Orders below a venue's minimum are simply rejected. Polymarket's is 5 shares, so the smallest position you can open depends on the price: about €1.00 at 20¢, €2.50 at 50¢, €4.75 at 95¢. Small covers and small experiments frequently fall under it.
Time
A share at 95¢ that resolves in twelve months returns 5.3% if it is right — and your money is unavailable for those twelve months. Nothing on the screen labels this as a cost, but it is one, and it is why long-dated favourites tend to trade below what their real probability would suggest.
How a market actually ends
A market does not resolve because the event happened. It resolves because a defined process concluded that its written resolution criteria were met. Those criteria are the real product: they specify the source, the deadline, the tie-breaks and what happens if the question becomes unanswerable. Two markets on what looks like the same question can settle differently, and both be correct.
On Polymarket, that process is an oracle with a dispute window: someone proposes the outcome, and there is a period in which it can be challenged before the result is final. In practice this means the last stretch is not instant — a market can sit at 99¢ for a while, decided in substance and not yet settled in money.
Then there is a step most people don't expect: resolving is not paying. Winning shares become redeemable, and redeeming is an action you take. On Polymarket the redemption pays out bridged USDC.e, which is a different token from the one the exchange trades against — so freshly claimed winnings can appear to be missing until they are converted. The markets themselves run on Polygon.
Three places "price = probability" isn't quite true
- Time has a price. Money in a long-dated market is money not earning anything else. A share at 95¢ resolving next week and a share at 95¢ resolving next year are not the same claim, and the market knows it.
- The mid is not tradable. You buy at the ask and sell at the bid. The probability you can act on is a band — 61 to 63 on the book above — not the 62 in the middle of it.
- Thin books carry thin information. A market at 2¢ with €40 resting behind it is a number, not a forecast. Depth is the difference between a price and an opinion.
And a fourth that isn't about markets at all: the price is the probability of the stated criteria being met. If you read the title and not the criteria, you may be trading a different question than the one you have a view on.
The venues differ in the mechanism, not the idea
Several venues run prediction markets, and the same question can trade at different prices on each. What changes your execution is not the branding but how the price is formed.
| Venue | How a price is formed | What that means for you |
|---|---|---|
| Polymarket | Central limit order book | The deepest catalogue and the ladder above. You can rest a limit order and be the one earning the spread. Minimum 5 shares; exits are redeem-then-convert. |
| Limitless | Order book | Same shape, generally shorter-dated markets and its own liquidity, so prices can diverge from Polymarket on the same question. |
| Myriad | Automated market maker | No book — you trade against a curve, so your own order moves the price as it fills. Size matters more than depth, and slippage is a setting rather than a surprise. |
| Azuro | Pooled liquidity, quoted per bet | Closer to a sportsbook: you are given a price for your stake and accept or decline it, with a minimum-odds guard so a moving price cancels instead of filling worse. |
Each venue runs on its own chain and settles in its own stablecoin, which is why moving between them is a transfer and not a click. Check the venue's own documentation for current chains, collateral and fee schedules before you fund anything — those change.
What goes wrong
- You traded a different question. The title matched; the resolution criteria didn't. This is the most common expensive mistake.
- Your order under-filled. A market order that cannot find enough resting size fills partly and cancels the rest. Read what landed, not what you submitted — a position half the size you intended is a different bet.
- You can't get out. Selling needs a buyer. In a market nobody is watching, the bid may be far below where the last trade printed, or absent entirely. Entry liquidity is not exit liquidity.
- The money is locked longer than the event. Between the event, the resolution process and your redemption, capital can sit idle well past the moment you were right.
- The paperwork is not the same as a bet's. Where you live, a position in a market may be treated as an asset disposal rather than a wager, with a different set of records required. Keep the trades, the prices and the dates.
Eligibility and risk. Prediction markets are for adults (18+) and are restricted or unavailable in a number of jurisdictions — including some where amparo itself is not offered, and availability is checked at sign-up. Prices move against you, markets resolve on their own stated terms, and money in an open position is money you cannot reach. This page is educational; it is not investment, betting or tax advice.
amparo puts several of these venues on one board with the depth visible before you commit, and caps how far a market order may walk. Reading the board is free; placing orders is the part that is paid for.