“Yes”, “No”, Web3Net: How decentralised prediction markets work

Prediction markets have existed online for roughly 30 years; in 2024 their Web3 incarnations took off amid America’s presidential race. Sergey Golubenko explains how an idea from the Austrian school morphed into a business—and into a theory of how to build a better society.

Early prototypes

Prediction markets are speculative venues where users trade contracts on the outcomes of future events via dynamic, open platforms.

A by-product of the mechanism is statistical information. It helps scholars and businesses understand correlations between social behaviour and various political and economic factors.

One of the first prediction markets was created in 1988 by a group of economists at the University of Iowa. The goal of the Iowa Electronic Markets project was to study the accuracy of collective forecasts for economic and political outcomes. The platform operates as a non-profit and caps stakes at $500.

As the internet matured, such platforms became more accessible and took the form of modern prediction markets. Among early projects, two stand out:

• Hollywood Stock Exchange, a cinema-market simulator launched in 1996. The game lets users trade “shares” (forecasts) and “options” on films, actors and directors. In 2006 the platform’s users correctly predicted 32 of 39 Oscar nominees and seven of eight winners. Clients included the studios MGM and Lionsgate;
• Hypermind, a prediction exchange active since 2000 where users wager virtual money on political, sporting and financial events. Multinationals were among the first to harness its big-data advantages for corporate use. Over the years its clients included Hewlett-Packard, Corning, Renault, Eli Lilly, Pfizer, Siemens, Masterfoods and Arcelor Mittal.

How prediction markets work

Web3 prediction markets structurally resemble classic gambling, but the operating asset is cryptocurrency. The terminal value of the tokens through which users place bets is linked to a specific event or parameter.

Topics range from Olympic results to climate change and presidential elections. Markets can be divided into several types by forecasting complexity, use case and the number of inputs:

• Binary. A choice between two outcomes, for example backing one of two presidential candidates;
• Categorical. Participants pick one among many slots, for instance which top-10 cryptocurrency by value will rank third in August 2025;
• Scalar. Results fall anywhere on a sliding scale under the logarithmic market scoring rules. Payouts vary with proximity to the realised outcome. One application is forecasting an asset’s price;
• Combinatorial. A large space of variable combinations. For example, which L2 network with the priciest transactions but the highest throughput will lead by TVL over a year.

Consider a fictional example about the potential dismissal of Gary Gensler: “Will the chair of the SEC leave his post before 1 February 2025?”

The most common binary markets offer two interchangeable tokens: “Yes” or “No”. These trade from 0% to 100%—priced from $0 to $1—reflecting the crowd’s consensus.

Those expecting a positive outcome buy the “Yes” token at $0.50, implying a 50% chance that Gensler will be removed. Those who think otherwise buy “No”.

If headline-grabbing news and rumours push more traders to expect his departure, the price of “Yes” will rise and “No” will fall. Should the prediction prove correct, “Yes” holders will see their tokens settle at $1 apiece, while “No” holders lose their stake.

Two main payout methods are common. In one, the potential winnings are fixed upfront, allowing precise reward calculations. In the other, payouts are parimutuel (totalisator): the prize pool is split among all winners, making it advantageous at times to bet against the crowd.

The wisdom of crowds

Prediction markets are used to gauge the probability of upcoming events. Proponents argue that shifting crowd sentiment can reveal why and how change happens.

A major theoretical contribution came from Friedrich von Hayek, a leading light of the Austrian school and Nobel laureate. In “The Use of Knowledge in Society” (1945) he argued that information in society is dispersed, and that markets aggregate it efficiently, enabling better decisions. His ideas underpinned the efficient-markets hypothesis and inspired designers of prediction markets to build platforms that tap “collective wisdom”.

In the early 2000s the American economist Robin Hanson proposed futarchy, a hypothetical governance system that sets a state’s primary goal as maximising social welfare. To achieve it, one defines clear, measurable targets: economic growth, population happiness and other indicators.

Hanson argues that when participants stake money or other real resources on prediction markets, they are motivated to make honest, considered forecasts. That should increase accuracy—and, in turn, governance efficacy.

The power of collective intelligence is captured in James Surowiecki’s bestseller “The Wisdom of Crowds”, which argues that well-designed prediction markets can surface knowledge embedded in disparate groups and improve forecasts compared with polling individual experts.

Decentralised prediction markets

With Web3, prediction markets run as dapps, gaining decentralisation, transparency and partial anonymity. Here are the sector’s main projects.

Polymarket

Polymarket, founded by American entrepreneur Shayne Coplan in 2020, is a decentralised platform for betting on existing events and creating new ones.

Participants trade outcome shares. The more users back a particular result, the higher its price. If they are right, they receive rewards proportional to their stake.

The platform offers anonymity by forgoing KYC, though it is not immune to future regulatory risk. In 2022 the CFTC fined the startup $1.4m for unregistered activity.

Polymarket runs on the Polygon L2 network with relatively low transaction costs; smart contracts ensure timely, secure settlement. To keep large, data-heavy markets functioning, it uses UMA’s Optimistic Oracle.

An AMM also lets users exit positions before resolution.

The platform’s popularity has surged since May 2024 amid the U.S. election. That month the project raised $70m across two rounds from General Catalyst, DragonFly Capital, ParaFi Capital, Vitalik Buterin and others.

According to Dune, alongside record July transaction volumes, more than 65,000 new users registered in a month—nearly 16 times April 2024’s tally. The project has no native token yet.

Gnosis

Launched in 2015, Gnosis leans into decentralisation and freedom via a DAO. In 2020 GnosisDAO was created to pursue futarchy-style governance.

The team developed Gnosis Conditional Tokens, designed for prediction markets and other condition-dependent applications. They split assets into scenarios, with each token representing an outcome tied to a specific event.

In April 2020 Gnosis introduced Corona Information Markets, where users could forecast aspects of the COVID-19 pandemic. In July the same year DXdao launched the decentralised prediction market Omen.eth.

The ecosystem also includes:

• Azuro Protocol (AZUR), a protocol for configuring third-party prediction markets;
• the Safe multisig wallet;
• DEX CoW Protocol;
• the EVM-compatible sidechain Gnosis Chain;
• the Gnosis Pay payment system;
• the GNO token.

Gnosis Chain stands out for its high degree of decentralisation and requires only 1 GNO to spin up a new validator. According to Dune, as of 23 August 2024 the number of provers exceeds 231,000.

Augur

One of the first Web3 prediction-market projects, Augur was founded in 2014 by Jack Peterson and Joey Krug. Vitalik Buterin advised the developer, Forecast Foundation.

In April 2015 the team deployed its first smart contract on Ethereum. Beta testing began in March 2016. Two years later the platform officially launched.

In July 2020 the Augur v2 protocol went live. A key change was settling markets in U.S. dollar equivalents using DAI. The REP token contract migrated to the new version.

In November 2021 Forecast Foundation joined DXdao, the organisation behind Omen.eth.

Other Web3 prediction-market projects include:

• Polkamarkets — operates on Ethereum, Moonbeam, Moonriver and Polygon, using USDT. Markets use AMMs to maintain liquidity, and liquidity providers receive a share of trading fees. Creating a market requires 1,000 POLK tokens;
• SX Network — a prediction platform launched in 2021. The SX token secures and governs SX Bet. SX Network is an EVM-compatible blockchain that works with Ethereum and Polygon. Since 2019 it has processed more than $400m in betting volume;
• Kleros (PNK) — a decentralised arbitration protocol for resolving disputes in the on-chain economy via prediction markets;
• PancakeSwap — a decentralised exchange that introduced a prediction market on Arbitrum powered by AI technology, the Allora Network model. Users can bet on crypto price moves every ten minutes; successful forecasters share a prize pool;
• Vega — a network backed by Coinbase Ventures that uses Cosmos infrastructure. It focuses on perpetual futures. In an update, the team said it will soon enter the prediction-market race.

Conclusions

Some markets suffer from thin participation, limiting the reliability of their signals for lack of liquidity. On AMM-based venues, slippage risk bears watching. Markets can be manipulated, and not every facet of society lends itself to quantitative prediction.