If you have watched prices on a regulated event contract or a crypto prediction platform, you have seen quotes like YES @ 0.37 or 42¢. The central claim of this chapter is simple: in a well-designed binary market, price ≈ probability—not because the exchange prints odds, but because traders with money on the line push price until it reflects their collective belief.
That link is the information aggregation thesis: markets turn private guesses, models, and news into one public number that updates in real time. Chapter 1.1 defined the contract; this chapter explains why the number on the ticket is treated as a forecast at all.
From price to implied probability
Think of a YES share as a lottery ticket that pays $1 if you are right. If you would happily pay up to $0.40 for that ticket, you are revealing a belief near 40%—plus whatever risk premium you require. That is the intuition behind mapping price to probability.
For a standard $1 binary that pays $1 if YES wins and $0 otherwise, a YES price of $0.10 suggests roughly a 10% chance, $0.50 suggests a coin flip, and $0.82 suggests about 82%. The NO side is the complement: if YES is $0.82, NO trades near $0.18, ignoring spread and fees. Traders arbitrage obvious gaps, so YES plus NO should sum to about $1 in liquid markets.
Price is a transaction price, not a survey answer. It embeds fees (taker fees, withdrawal costs, on-chain gas), the bid–ask spread (you might pay 41¢ to buy YES while sellers only receive 39¢), time value and risk preference on long-dated or tail events, and a liquidity premium when thin books exaggerate moves. Treat “price = probability” as a working model, then refine when spreads are wide or rules are exotic.
Why trading produces a forecast
Imagine a market on “Rate cut at the September meeting.” Before the announcement, a macro analyst buys YES because her model says 70%. A hedger sells YES because his firm already benefits from high rates. A journalist buys YES after a sourced leak. A market maker supplies liquidity on both sides.
Each order shifts price. When the marginal buyer will only pay $0.64 and the marginal seller will only accept $0.66, the mid-price $0.65 is where disagreement clears. That mid is the market’s consensus odds at that second.
This is Hayekian in spirit: no central planner collects forecasts; participants reveal information through trades. The thesis says that process can outperform static polls when traders are incentivized, liquidity is deep enough to absorb informed flow, and resolution is objective and trusted. When those fail, price is still a number—but it may track hype, manipulation, or a tiny clique of traders.
Order flow as news
Prices move on order flow, not press releases alone. A headline matters only insofar as it changes what people will pay. The usual sequence runs from news to belief update to orders to price change to others reacting to a new equilibrium. Fast traders react in seconds; slower participants join over hours. That is why prediction markets can front-run stale polls: the poll is a snapshot; the market is a live auction.
Aggressive buying of YES pushes price up and implies higher probability. Large selling of YES (or buying NO) pushes price down. Watching who trades is often impossible on anonymous accounts, but size and speed still signal conviction.
Efficient prices are conditional
“Efficient” does not mean “correct forever.” It means hard to beat after fees given public information. A deep two-sided book produces smooth prices and more trustworthy implied percentages. A whale dominating volume makes price the whale’s view plus noise. Manipulation or wash trading creates fake signal. Ambiguous resolution makes traders price legal risk, not just the event. Restricted access (US-only, KYC) shrinks and may bias the participant pool.
A $200k political market on a major platform behaves differently from a $2k meme market on-chain. Always read liquidity alongside price.
Worked example: repricing after new information
Suppose YES on “Bill passes committee this week” trades at $0.30 (about 30% implied). A credible leak suggests whip counts have shifted. Informed traders buy YES; price rises to $0.45, then $0.52 as liquidity providers adjust quotes. Holders who bought at 30¢ can sell at 52¢ for profit before the vote—classic price discovery.
If the bill fails, YES settles to $0. The market was “wrong” ex post but may have been reasonable ex ante at 52¢—forecasting is probabilistic, not binary grading.
NO prices and multi-outcome markets
On a binary market, focusing only on YES is enough: P(NO) ≈ 1 − P(YES).
Multi-outcome events (three or more candidates) require prices on each outcome that should sum to roughly 100%, again minus spread and fees. If Candidate A is 40¢, B is 35¢, and C is 20¢, the book implies 95% total—the missing slice is often spread or an “Other” bucket. Mispricings appear when one outcome is illiquid, when correlated bets exist across related markets (“wins nomination” vs “wins general”), or when traders arb across platforms with different rules.
Information aggregation vs. manipulation
The same mechanism that aggregates truth can aggregate lies if manipulators cheaply move thin markets. Defenses include liquidity incentives and market maker programs, position limits and surveillance, longer trading windows so informed capital can counteract pumps, and clear resolution to reduce legal arbitrage.
Regulators worry about manipulation around elections; practitioners watch for suspicious flow—new wallets, size at odd hours, correlated markets not moving. A price spike without corroborating news in related contracts is a red flag.
Fees as a hidden probability tax
A 2% taker fee does not change the headline price on the tape, but it changes the breakeven probability you need to profit. If you buy YES at 50¢ and pay 2% on entry and exit, you need a higher true chance than 50% to expect positive value. Professional traders bake fees into their models before calling a market “mispriced.”
Comparing to polls and models
Polls update when a field period ends; their strength is representative sampling when methodology is sound. Fundamentals models update when an analyst reruns a spreadsheet; they bring structure and economic priors. Prediction markets update on each meaningful trade; they compress many inputs, including polls, into a tradable price.
Markets do not replace polls—they synthesize them under incentives. A smart trader buys YES when her model says 60% but the market says 45%.
Mid-price, last trade, and what you actually pay
Platforms display different numbers. The last trade is the most recent matched price. The mid is halfway between the best bid and best ask. When you buy aggressively, you pay the ask, which is above the mid; when you sell, you hit the bid, below the mid.
For quick implied-probability reading, mid is fine in deep markets. For sizing a trade, use the side you will actually hit. A market “at 50%” with a 45¢ bid and 55¢ ask is really two different economics for buyer and seller.
Time to resolution and drift
Long-dated contracts can sit at odd levels because few traders want to tie up capital for months. Near expiry, prices often snap toward 0 or 1 as uncertainty collapses—unless resolution is disputed. A contract at 70% three days before a vote is not the same animal as 70% six months out; the latter embeds more room for surprises and more illiquidity risk.
How polls enter the price (without replacing it)
Sophisticated participants often run explicit models that ingest poll averages, economic indicators, and fundamentals, then trade when the market diverges. That means poll information may already be inside the price by the time you read it on Twitter. The edge is not “I saw a poll”—it is “I understand why the market is still at 45% after that poll.”
How to read a live quote
When you see a live quote, ask what contract pays $1 on (exact resolution text), what the spread is (wide spread means fuzzy implied %), how much volume or open interest exists (thin books deserve skepticism), when trading ends (prices can drift near expiry), and whether related markets are consistent (arb gaps mean opportunity or rule mismatch).
If you are writing for an audience, round aggressively when liquidity is weak. Saying “low forties with a wide spread” is more honest than “41.3%” on a thin book.
Connection to crowd wisdom
The information aggregation thesis is the mechanical half of the story; the next chapter asks whether the people behind the orders are diverse and independent enough for that mechanism to work. Price equals probability only when the auction is real.
Key takeaways
Binary YES price near $p means the market roughly prices YES at p%, subject to fees and spread. Prices change because orders change; news matters through trading. The information aggregation thesis claims decentralized, incentivized trading produces useful public forecasts when liquidity and rules are sound. Efficiency is conditional—always cross-check liquidity, resolution, and participant access.
Next: The Wisdom of Crowds: Why Groups Are Smarter Than Experts