Opinion Markets Like Polymarket: Do They Actually Work?

Opinion markets - platforms like Polymarket and Kalshi where you buy and sell contracts on whether an event will happen - work remarkably well as information machines and poorly as wealth-building tools, and both halves of that sentence are supported by evidence. By April 2026 the two largest platforms were trading about $24 billion a month combined, up from under $5 billion in September 2025, and the US regulator proposed formal rules for the sector in June 2026. In India, meanwhile, homegrown "opinion trading" apps were banned under the Online Gaming Act of 2025. This guide examines how these markets work and where they break, using three lenses: the economics of information, the psychology of money, and the principles that separate investing from betting.

What is an opinion market and how does the price work?

A prediction market sells contracts that pay a fixed amount, typically $1, if an event happens and nothing if it does not. The price at any moment is the crowd's implied probability: a "Yes" contract trading at 65 cents means the market collectively assigns a 65% chance to the event.

If you believe the true probability is higher, you buy; if lower, you sell. Every trade nudges the price towards the beliefs of whoever is willing to back their view with money. When the event resolves, the contract settles at $1 or $0, so unlike a stock there is no ambiguity about final value and no long run to wait for. Kalshi runs a US-regulated exchange; Polymarket historically ran offshore on crypto rails and, as of 2026, also owns a smaller US-regulated venue. In June 2026 the CFTC, the US derivatives regulator, proposed rules that would permit most event contracts while banning those most vulnerable to manipulation, such as bets on individual player injuries or referee decisions.

Why do economists take these markets seriously?

Because they are a working demonstration of two old ideas: Hayek's knowledge problem and the wisdom of crowds. Hayek argued in 1945 that the information needed to set correct prices is scattered across millions of people and can never be centralised; markets are the mechanism that aggregates it. A prediction market does this in its purest form - the "asset" is literally a probability.

The wisdom-of-crowds result, popularised from Francis Galton's 1906 observation that a fair crowd's average guess of an ox's weight beat individual experts, adds the statistical logic: when many independent estimates are averaged, individual errors cancel. Prediction markets improve on a simple average in one respect - participants are weighted by conviction, because backing a view with money filters out cheap talk. This is the "skin in the game" property: a poll asks what you hope, a market asks what you will pay for.

The empirical record is real but conditional. In the 2024 US presidential race, the major platforms identified the winner earlier and with more confidence than most polling aggregates. But a study of the same election cycle found accuracy varied sharply by platform and liquidity: roughly 93% of PredictIt markets beat chance, against 78% on Kalshi and 67% on Polymarket, with the largest errors in thin, low-volume markets. The machine works when its conditions are met and degrades when they are not.

When does the wisdom of crowds break down?

Crowd forecasts are only wise when estimates are diverse, independent, and numerous - and opinion markets routinely violate all three. This is the central critical point, and it comes straight from the theory's own fine print.

• Independence fails. Traders watch the same news, the same influencers, and the market price itself. When everyone updates from the same signal, errors stop cancelling and start compounding - an information cascade, not aggregation.
• Diversity fails in thin markets. A market on a niche question may have a handful of traders with similar views. The 2024 accuracy data showed exactly this: errors concentrated where liquidity was lowest.
• Whales overwhelm the crowd. In October 2024, a single French trader using eleven linked accounts bet over $28 million on a Trump victory and pushed Polymarket's implied probability 10 to 15 percentage points above rival platforms for weeks. He won about $85 million, and researchers still debate whether that was superior private information (he had commissioned his own polls) or a demonstration that one conviction-heavy actor can move the "crowd's" answer.
• Resolution ambiguity. Contracts settle on written criteria, and disputes over what counts as "the event happening" have repeatedly caused controversial settlements. A probability is only as trustworthy as the referee who decides the outcome.

Keynes described professional speculation as a newspaper beauty contest where you win not by picking the prettiest face but by picking the face others will pick. Near resolution, prediction markets escape this because reality itself settles the bet. Far from resolution, though, prices can drift on what traders think other traders think - probability as fashion.

What does money psychology say about the people trading them?

Behavioural economics predicts specific, measurable distortions in these markets, and the data confirms them. The best documented is the favourite-longshot bias: bettors systematically overpay for low-probability outcomes and slightly underpay for near-certainties. Studies of betting and event markets, including NBER work testing whether the cause is risk-love or misperception, find that cheap "longshot" contracts win far less often than their price implies, while expensive favourites win more often - meaning the exciting 5-cent moonshot is, on average, the worst-priced item on the platform.

This is prospect theory in action: people overweight small probabilities, which is also why lotteries exist. Three other forces do the remaining damage. Overconfidence - most traders believe their political or sports judgment is above average, and a market needs exactly such people to supply the losing side of trades. The illusion of control - forming an opinion feels like skill, so opinion trading feels like investing even when the payoff structure is a coin flip you paid a fee to enter. And gamification - fast resolution, streaks, leaderboards and constant near-misses deliver the variable-reward loop that keeps engagement high, the same loop that makes F&O trading feel like gambling for retail participants.

There is also a quieter economic puzzle: the no-trade theorem. If everyone were rational and traded only on information, hardly anyone would trade at all - because the moment someone offers you a bet, you should ask what they know that you do not. Real volume exists precisely because a large share of participants trade for entertainment or conviction rather than edge. In plain terms: the market's forecasting accuracy is subsidised by the losses of its recreational users.

Are opinion markets investing, by any standard principle?

Measured against the basic principles of investing, event contracts fail on structure, not on morality. The differences are mechanical:

1. No productive asset. A share is a claim on future cash flows that can grow; the pie expands, so all holders can win over time. An event contract redistributes a fixed pot. Before fees it is zero-sum; after fees and spreads it is negative-sum for participants as a group.
2. No compounding. Contracts expire at $1 or $0. There is no dividend, no reinvestment, no long horizon to repair a mistake. Time, the investor's main ally, is absent.
3. No diversification within a bet. A binary contract has exactly two outcomes. Position sizing rules like the Kelly criterion show that even with a genuine edge, betting large fractions of capital on binary outcomes produces violent drawdowns - and most users have no edge to size.
4. Edge is provable and rare. The consistent winners documented on these platforms are market-makers earning spreads and quantitative traders exploiting biases like the longshot effect - the same professional-versus-retail structure seen everywhere else in speculation.

None of this makes prediction markets useless. It makes them what they are: a forecasting utility with a trading interface, where the social value (better public probabilities) is a by-product of individually negative-expectation activity for most participants.

What happened to opinion trading apps in India?

India's version of this industry was built on smaller stakes and faster questions - "Will it rain in Mumbai tomorrow?", "Will India win the toss?" - and it ended in a ban. The Promotion and Regulation of Online Gaming Act, 2025 prohibited online real-money games, and leading opinion trading platform Probo discontinued its real-money operations, having also faced Enforcement Directorate searches under money laundering law and FIRs under public gambling statutes.

The Indian episode is a clean case study in the regulatory question every country is now facing: the same contract can be framed as information discovery (the US CFTC's view, with guardrails) or as gambling by another name (India's view, at least for retail money apps). Both framings are describing real properties of the same instrument.

So do opinion markets work or not?

As forecasting machines: yes, conditionally - liquid, high-attention markets close to resolution produce probabilities that are hard for polls or pundits to beat. As a way for a typical participant to grow money: no, structurally - negative-sum design, binary payoffs, documented behavioural biases, and professional counterparties see to that. Both findings can be true at once because the market's accuracy does not require its average user to profit; it only requires them to pay for their opinions.

What this means for you, as education: prices on these platforms are worth reading the way you would read any good forecast - a fast, incentive-weighted summary of public belief, best trusted when volume is deep and the question is crisp. The psychology chapter is the more personally useful one: the pull you feel towards a 5-cent longshot, the confidence you feel in your own opinion, and the fun of fast resolution are precisely the mechanisms the structure monetises. Recognising them there makes them easier to recognise in stock tips, IPO frenzies and crypto seasons too.