Reading prices as implied probability, the number behind the number

Why the price already is the percentage

Almost every contract on a regulated event exchange is built the same way. It pays exactly 1 dollar if the stated outcome happens and exactly zero if it does not. That single design choice is what lets you skip the maths. If a contract that pays 1 dollar is changing hands at 37 cents, the people buying and selling it are collectively saying the dollar is worth 37 cents to them right now. The only reason a future dollar is worth 37 cents today is that they judge its chance of arriving at roughly 37 percent. Move the decimal point and you have the probability. Kalshi states this plainly in its own guidance: a yes price of p cents implies a market estimate of about p percent that the event happens, and about 100 minus p percent that it does not, per Kalshi educational material, as of June 2026.

The no side is simply the rest of the dollar. If yes trades at 37 cents, no trades at about 63 cents, because someone holding the no contract collects the dollar in the other 63 percent of cases. This is why the two prices sit on opposite ends of the same dollar. When you see yes at 37 and no at 63 you are looking at one probability described from two directions, not two separate opinions. A useful habit is to glance at both and check they add up close to a dollar. If they do, the market is internally consistent. If they are far apart, something is off, and the section below explains what.

This is also why a price carries information that a simple headline does not. A poll might say a candidate is ahead. A model might say an outcome is likely. A market price says how likely, in a single number that updates the moment new information arrives and that costs real money to be wrong about. That last part matters. The people setting the price are not answering a survey, they are putting capital behind their estimate, which is the mechanism that is meant to keep the number honest.

The spread and the fee take their cut

An order book has two live prices at once. There is the best price someone will pay to buy, called the bid, and the best price someone will accept to sell, called the ask. They are never exactly equal. The gap between them is the spread, and it exists because buyers and sellers each want a little edge. On a busy market with many participants the spread might be a single cent. On a quiet one it can be five cents or more. The true implied probability is not the bid and not the ask, it is the midpoint between them. Reading the midpoint, rather than whichever side last printed, is the difference between a soft estimate and a sharp one.

This is also why yes and no often add up to a touch more than a dollar rather than exactly a dollar. If you could buy yes at its ask and no at its ask, you would be paying both spreads, so the two asks sum to slightly above 1 dollar. The overshoot is the cost of crossing the market on both sides. A wide combined total is a quiet, thin market where the implied probability should be treated as approximate. A total sitting right on a dollar is a liquid market where the number is as crisp as the venue allows.

Then there is the fee. A fee does not change the market estimate of probability, but it changes the probability you personally need for a position to be worth holding. Take Kalshi as the worked example because it publishes a clear formula. Its taker fee is 7 cents multiplied by the price, multiplied by one minus the price, charged per contract, which works out to a maximum of about 1.75 cents per contract at the 50 cent level and less toward the edges, per the Kalshi fee schedule effective February 2026. Maker orders, the resting limit orders that wait on the book rather than crossing it, are charged at one quarter of that, with a cap of 3.5 cents per contract. The exact numbers differ by venue, and several platforms fold their cost into a wider spread instead of a named fee, but the principle is universal. Whatever you pay to enter is added to the price when you work out the true chance you need just to come out level. We track these costs side by side in the cross platform fees dataset, refreshed on a weekly cycle.

One price, read five ways

The table below takes a single yes price and translates it into the things you actually want to know: the implied chance, what a winning contract returns, the profit on a one dollar contract, the equivalent decimal odds, and the true chance you would need just to break even once a representative fee is included. It shows why a cheap contract is not a bargain and an expensive one is not a trap. Each is simply a different probability described in a different unit.

Methodology: implied chance is the price read across; decimal odds are 1 divided by the price; profit assumes a contract bought at the listed price and settling at 1 dollar. The breakeven chance adds a representative Kalshi taker fee of 7 cents times price times one minus price per contract to the cost, per the Kalshi fee schedule effective February 2026. Figures are rounded and illustrative. Fees, spreads, and contract terms vary by platform and change. Verify the current schedule on each venue before relying on it.

Reading a field of outcomes

Plenty of markets are not a simple yes or no. A question such as which of several candidates wins, or which range an economic figure lands in, has many possible answers, each listed with its own yes price. The reading rule does not change. Each line still pays 1 dollar if that answer is the one that happens, so each price still reads across as the chance of that single answer. What changes is that the prices for all the mutually exclusive answers should add up to roughly 100 percent, because exactly one of them will be true.

That total gives you a quick health check. If the listed answers sum to a long way above 100 percent, the market is charging a wide combined spread and the individual numbers are soft. If they sum below 100 percent, either the list is incomplete and an other or none answer is missing, or there is a genuine pricing gap. A careful reader normalises the field in their head: take one answer priced at 30 cents in a market whose lines sum to 108, and its honest share of the probability is a little under 30 percent once you divide out the overcounting. The arithmetic is rough, but it stops you reading a crowded field as more confident than it is.

The same idea explains why a favourite in a large field can look cheap. If twelve answers each carry some chance, even the leading one may trade well below 50 cents, not because the crowd is unsure it leads, but because the remaining probability is spread thinly across eleven others. Read the leader against the field, not against a coin flip.

Reading one market page end to end

Imagine a contract on whether a named economic figure comes in above a threshold, with yes showing a best buy of 61 cents and a best sell of 63 cents, and a last traded price of 58 cents printed two hours ago. Start with the midpoint. The live estimate is 62 percent, not the 58 percent the stale last print suggests. The gap between 58 and 62 is exactly the trap of reading the last print in a slow market, and it is four whole points of probability in this example.

Now widen the view. The no side shows a midpoint near 38 cents, so yes and no sum to about 100, which tells you the book is reasonably tight and the 62 percent reading is trustworthy as far as it goes. Next add the cost of acting. Buying yes at the 63 cent ask and paying a representative fee near a cent and a half means the figure has to clear the threshold a little more than 64 percent of the time for the position simply to come out level. That is the difference between what the market thinks, 62 percent, and what you would need to be true, a bit above 64 percent, just to break even.

Finally, place the number in time. If the figure is released tomorrow, the 62 percent is a sharp, near term estimate worth weighing seriously against your own read of the data. If release is months away, treat it as a softer number that the evidence says is likely compressed toward the middle. Nothing in this walk through tells you what to do. It tells you what the page is actually saying, which is the only honest starting point for any decision you then make for yourself.

Implied is not the same as true

A price tells you what the market believes. It does not tell you what will happen, and it does not even reliably tell you the real probability. The good news is that the evidence is broadly kind to these markets. Reviews of prediction market accuracy find that prices track realised frequencies fairly well and tend to edge out bookmaker odds, and that they sharpen as a market nears its close, when uncertainty has been resolved and money has concentrated the estimate. A contract at 80 cents in the final hours before settlement is a more trustworthy 80 percent than the same price months out.

The less comfortable news is the pattern researchers call the favourite longshot bias. Across many betting and prediction markets, very cheap contracts on unlikely outcomes tend to win less often than their price implies, while heavy favourites tend to win a little more often than theirs. In plain terms, longshots are usually a touch overpriced and favourites a touch underpriced. Academic work on these markets, including studies spanning large samples of contracts, has documented the effect repeatedly, and it is one reason a string of cheap contracts that each look like a small flutter can quietly lose money over time. Knowing the bias exists is part of reading the number without being fooled by it.

There is a horizon effect on top of that. Calibration is strongest when the event is close. For events far in the future, prices compress toward the middle of the scale, so a market sitting at 60 percent a year out may be expressing more uncertainty than genuine conviction. This is contested in its details and the size of the effect varies by category, so the safe reading is directional: trust a near term price more than a far term one, and treat anything resolving a long way off as a wide band rather than a fine point. None of this means the number is useless. It means the number is an estimate with known habits, and the skilled reader keeps those habits in mind. For scale, the CFTC noted that volume across its designated contract markets exceeded 25 billion dollars in 2025, so these are deep enough markets to take seriously while still being young enough to misprice.

It helps to remember why the number tends to be good at all. A market price is a weighted vote in which the people most willing to back their view with money count for most. When someone believes the price is too low they buy, which lifts it, and when they believe it is too high they sell, which lowers it. The price settles where those pressures balance, which is meant to be the level at which the marginal trader is indifferent. That mechanism is the reason a market estimate often beats a single expert or a raw poll average, and it is also the reason a thin market with few participants is less reliable, because there are simply fewer informed votes holding the number in place. Reading a price well means respecting the mechanism where it is strong and discounting it where it is thin.

Four traps when reading a price

Reading the number with judgement

Once a price reads as a probability, you can hold it next to other evidence. A market at 30 percent on an outcome that the latest polling, a published model, or official data all put nearer 45 percent is telling you the crowd and those sources disagree. That disagreement is interesting, and it is worth understanding before you do anything, but it is not a signal to act. Sometimes the market is slow and the model is right. Often the market has already absorbed something the model has not. The point of reading the number is comprehension, not a shortcut to a decision.

The most useful frame is to treat the price as a starting question rather than an answer. Why does the market think this is 30 percent? What would have to be true for it to be 45 percent instead? Where is the contract resolving from, and could the spread be hiding a thin book? A reader who can answer those questions understands the contract. A reader who simply spots a number they like and treats it as a tip understands nothing and carries all the risk. A price is an estimate, not a forecast, and certainly not advice. Markets are often wrong, and reading the number well includes always knowing how uncertain it is.

See how prices, spreads, and fees differ across platforms

Reading a price is the same skill everywhere, but the spread you cross and the fee you pay are not. Availability and the rules depend on where you live, so check legality first. These reference pages are information, not an introduction to any one platform.

Common questions, answered plainly