What Does the Price Mean in a Prediction Market?

In a prediction market, every contract works the same way: it pays $1 if something happens, and $0 if it doesn't. So the price of that contract (say, 65¢) is what the crowd is collectively willing to pay for that $1 outcome. Which means it's also their best collective guess at the probability: 65¢ implies a 65% chance.

That's what "prediction market price" means. Everything else in this article is explaining why it's more complicated than that one sentence suggests.

What the price actually is

Every standard prediction market contract works on the same basic logic: the contract pays $1 if the outcome happens, and $0 if it doesn't.

That's it. That's the whole structure.

So if a YES contract is trading at 65¢, the price is telling you that the market collectively believes there's roughly a 65% chance of that outcome occurring. You pay 65¢ now for a contract that might be worth $1 later, or nothing.

The math of any trade follows from the price:

Buy YES at 65¢ → you risk 65¢ to win 35¢ (if the event happens, you collect $1; subtract your 65¢ cost, net gain is 35¢)

Buy NO at 35¢ → you risk 35¢ to win 65¢ (if the event doesn't happen, you collect $1; subtract your 35¢ cost, net gain is 65¢)

The platform doesn't set these numbers. Traders do, through buying and selling, the same way stock prices form. When more people buy YES, the price goes up. When more people sell, it drops. The price is where the last buyer and seller agreed.

This is what "implied probability" means: not a calculation from a model, but the price the market has settled on, expressed as a percentage.

What price are you actually looking at?

The number on screen is usually the midpoint of two prices, not what you'd actually pay to trade right now. Most explainers skip this entirely, and it matters more than people realize.

When you look at a prediction market platform and see "67%" or "0.67" next to an outcome — whether you're reading Polymarket prices, Kalshi prices, or any other platform — that number is usually not the price of the last trade. It's not necessarily a price you can trade at right now, either. It's a number called the midpoint, the average of two other numbers that are doing all the real work behind the scenes.

Those two numbers are the bid and the ask.

The bid is the highest price any buyer is currently willing to pay for YES shares.

The ask is the lowest price any seller is currently willing to accept.

Say the bid is 64¢ and the ask is 70¢. The midpoint (64 + 70) / 2 is 67¢. That's what you see displayed as "the probability."

But if you wanted to actually buy YES shares right now, you'd pay 70¢ (the ask, because that's what sellers are offering). If you wanted to sell, you'd get 64¢ (the bid). The displayed 67¢ is a snapshot of where the market is sitting, not a price you can immediately execute.

Polymarket's own documentation makes this explicit: when reading Polymarket prices, the probability shown is the midpoint of the bid-ask spread. If that spread widens beyond $0.10, the platform switches to showing the last traded price instead.

Why does this matter? In busy, liquid prediction markets — think the presidential election or the Super Bowl — spreads are tight. Bid and ask might be 66¢ and 68¢, making the midpoint close to what you'd actually pay. But in a quieter market, spreads can be wide. A displayed probability of 67% in a market with a $0.20 spread is much fuzzier than a displayed probability of 67% in a market with a $0.02 spread. Same number on screen; very different signal quality.

Multi-outcome prediction market prices

So far we've been talking about binary markets — one question, two sides. But plenty of prediction markets have more than two outcomes.

Take Polymarket's "Which party will control the Senate?" market. Five possible outcomes. Each one has its own price. The logic stays the same: each price represents the implied probability for that specific outcome, but now there's an added constraint. In a well-designed market, all the prices should add up to roughly $1, because one outcome has to happen.

They often don't add up to exactly $1 in practice. Fees, spreads, and thin liquidity on some outcomes all create small gaps. A market where all five outcomes sum to $1.08 isn't broken; it's showing you the friction in the system. Traders who spot these gaps will buy the underpriced outcomes and sell the overpriced ones to lock in a risk-free return — a practice called arbitrage. For most readers, it's a useful reminder that "implied probability" is an approximation, not a mathematically precise calculation.

How prediction market resolution affects price

Before you trade any market, read the resolution rules — not just the title. The price you're buying into is a bet on how the rules interpret the outcome, not just on the outcome itself.

The price floats, moves, responds to news, gets pushed around by big trades, and then one day it stops. The market resolves. Resolution is the moment the official outcome is confirmed and the platform finalizes the result. Winning contracts pay $1, losing contracts pay $0, and trading stops permanently.

But there's something important in how that finalization works: the market title is the question, but the resolution rules are what actually determines who gets paid.

This sounds like a technicality until it isn't. Every platform specifies, in the fine print of each market, exactly what source will be used to determine the outcome, what counts as official confirmation, and what happens if the situation is ambiguous. A market titled "Will X happen by June?" might resolve YES, NO, or void entirely if the situation falls into an edge case the rules don't clearly cover.

Kalshi users have encountered exactly this: a prediction that turned out to be "correct" in the plain-English sense of the question, but resolved unexpectedly because of how the underlying rules handled a specific edge case. Ambiguity in those rules has cost traders money.

Should you trust prediction market prices?

The honest answer is: more than you'd trust a pundit, less than you'd trust gravity.

Prediction market prices earn their credibility from incentives. People putting real money behind a belief have a reason to think carefully. There's no reward for sounding smart if you're actually wrong. That doesn't mean every trader is rational. But bad bets tend to get corrected by traders willing to take the other side. That structure tends to push prices toward accuracy over time.

But "tends toward accuracy" is not the same as "is always accurate," and the research is clear on the places where prices drift.

Favourite-longshot bias is one of the most consistent findings. A study by economists Constantin Bürgi, Wanying Deng, and Karl Whelan analyzing over 300,000 Kalshi contracts found that cheap contracts — say a 10¢ YES — win less often than their price implies. Longshots are systematically overpriced. Expensive contracts, meanwhile, win slightly more often than their price suggests. In plain terms: if you consistently bet on unlikely outcomes because "10¢ seems cheap," you'll lose more money than the price suggests you should. The math is quietly working against you.

The hidden disagreement problem goes deeper. Economist Charles Manski showed that a market price tells you less than it appears to. Consider two scenarios: in one, nearly every trader agrees the probability is around 65%. In the other, half the market thinks it's 90% likely and the other half thinks it's 40%, and the trading has averaged out to 65¢. The price looks identical in both cases. But those are very different situations: one is genuine consensus, the other is a market split down the middle. The price alone can't tell you which one you're looking at.

Manipulation is real, if limited. Lab experiments found that when traders tried to push prices in a direction, other traders spotted the opportunity and bet against them, correcting the price back toward consensus. But a large-scale 2025 study found something different: randomly shocking prices across 817 real markets, effects were still visible 60 days later, with only partial correction. Smaller, less liquid markets are more vulnerable.

Prediction market prices are reliable the way a well-calibrated instrument is reliable: useful signal, known limits, not a direct readout of truth.

Six things beginners get wrong about prediction market prices

1. "65¢ means a 65% guarantee." It means roughly 65% of the time, if the market is well-calibrated, the event should happen. A 90¢ contract still loses around 1 in 10 times. High probability is not certainty.

2. "YES and NO prices always add up to $1." Designed to. In practice, bid-ask spreads mean the two sides don't form a clean $1 sum at any given instant. The gap is usually small, but it exists.

3. "If I'm right about the outcome, I made a good trade." Not necessarily. Whether a trade was good depends on your entry price versus the probability you should have assigned at the time, not just on the final result. As Kalshi's own educational content puts it: "You could win on a coin flip that you paid 60 cents on the dollar, but it was still a bad bet."

4. "The displayed price is what I can trade at." Usually it's the midpoint. The price you'd actually pay to buy is the ask (higher than midpoint). The price you'd receive if you sell is the bid (lower than midpoint). In liquid markets the gap is tiny. In thin markets it can be significant.

5. "Resolution is obvious, so the rules don't matter." The title poses the question; the resolution rules determine the answer. Ambiguity in those rules has cost traders money. Read them.

6. "It's the wisdom of crowds." It's the wisdom of whoever is trading there, under those rules, willing to put real money in. Who participates shapes what the price reflects. A market dominated by one large trader, or one where most participants share the same blind spot, can show a price that looks like consensus but isn't.

Prediction market prices are one of the more honest signals in public life: real money, real stakes, real-time updates. But like any signal, they reward the people who understand what they're actually measuring.