16 min readDecentralised prediction markets were a key trend in the crypto industry last year, predominantly because they outperformed pollsters when forecasting the winner of the 2024 US presidential election. Bets on Polymarket suggesting Donald Trump had a 58% chance of winning were closer to the outcome than most polls.
This article explains how prediction markets work before exploring their socio-economic impact and the risks users face.
The evolution of prediction markets
Prediction markets date back to betting on papal elections in the early 16th century, but the Iowa Political Stock Market (IPSM), later renamed Iowa Electronic Markets, was one of the earliest platforms that allowed users to trade financial instruments tied to event outcomes. Operated by the University of Iowa as a research project, the IPSM accurately predicted the result of the 1988 US presidential election, won by Republican candidate George H.W. Bush.
Decentralised prediction markets are a natural evolution of the concept, leveraging blockchain technology to enhance the user experience:
Trust- transactions and outcomes are recorded on immutable and publicly-visible databases, lowering the risk of fraud or manipulation.
Fairness- market forces dictate the odds rather than central authorities, which tends to lead to better payouts.
Lower costs- the peer-to-peer model means that users don’t have to pay fees charged by intermediaries.
Augur is a pioneer in the sector. Launched in 2018 and built on Ethereum, it offers three types of markets: binary (only two possible outcomes), multiple choice and scalar (which involves choosing a price in a range). Users can purchase shares in ether or the DAI stablecoin, and they receive rewards in the form of REP, Augur’s native token. Holders of REP can stake it to participate in outcome verification or to raise a dispute.
How do decentralized prediction markets work?
Decentralised prediction markets sell tokens representing different event outcomes, which users purchase in crypto. The higher the price of the token, the more likely the outcome, and vice versa. When the event ends, holders of the winning tokens receive a payout. Users can also trade the tokens if the probability of their outcome changes.
The terms of the bet are recorded on a smart contract, a computer program that automatically executes when predetermined conditions are met. However, smart contracts can’t interact with the real world, so they need a way to confirm the outcome. That’s where oracles come in, networks that serve as a bridge between blockchains and off-chain data. They gather data and relay it to the smart contract so it can reward the winners. Oracles are either centralised, meaning they’re controlled by a single entity and rely on one source, or decentralised, which connect to multiple sources.
Decentralised prediction markets use a trading mechanism called automated market makers (AMMs), which are pools of tokens managed by smart contracts. These pools contain pairs of tokens, and every time a trade occurs, the weighting and price change, reflecting the market’s view of the outcome. The main benefit of AMMs is ensuring liquidity, but smaller pools may experience slippage, where the price shifts while an order is processed. AMMs are also subject to fluctuations in gas (transaction) fees charged by the underlying blockchain.
Payouts can be fixed or variable. Fixed payouts involve users buying tokens that represent a binary outcome. They’re priced between $0 and $1, with the price indicating which is more likely. So if a token costs 60c, that outcome has a 60% probability of success. When the event concludes, users holding winning tokens receive $1, while the others get nothing. Fixed payouts are simple to understand and encourage continual price discovery, but they rely on high liquidity.
Variable payouts are known as pari-mutuel. All bets are pooled, and the winners receive a share of the pool proportional to the number of tokens they hold. This mechanism is fairer, but it’s complex and the exact payout is only confirmed after the event concludes.
The process of confirming the outcome varies from platform to platform. Augur calls it the ‘reporting’ phase, which typically lasts a day. An initial reporter, who has staked REP tokens, assigns a ‘tentative winning outcome’.
The graphic below demonstrates the next steps:
Decentralized prediction markets: what kind of socio-economic impact?
Decentralised prediction markets are more than a way to earn rewards; they also serve as a reliable indicator of public sentiment about an event. While pollsters had the recent US presidential election neck and neck up until the last minute, Polymarket’s forecast of a Trump victory was more accurate.
One of the reasons that prediction markets are relevant is they offer a tangible, financial incentive. The possibility of losing money encourages users to provide reliable projections, as opposed to voters who don’t suffer any consequences if polls are wrong. Personal bias is less influential, whereas ‘shy Trump voters’, reluctant to reveal their political affiliation to researchers, are a possible explanation for pollsters’ poor performance during the 2024 US presidential election.
Prediction markets are also available to trade around the clock, unlike polls, which are conducted periodically and represent public views at a point in time. They provide live forecasts for the outcome of an event, allowing users to respond to new data, for instance when Donald Trump’s support fell after Kamala Harris entered the presidential race.
However, they aren’t foolproof- Hillary Clinton was the strong favourite to win the 2016 US presidential election, but she lost to Donald Trump.
Regulatory scrutiny of decentralised prediction markets
Founded in 2020 and built on Polygon, Polymarket is one of the biggest decentralised prediction markets, ending 2024 with $9 billion in total trading volume and over 300,000 users. The 2024 US presidential election alone attracted bets worth $3.2 billion, while Bloomberg reported that several Wall Street firms were monitoring the site’s odds, including investment bank JP Morgan.
Despite its success—or possibly because of it—Polymarket has been targeted by U.S. financial authorities. It settled a $1.4 million fine with the CFTC in 2021 and has been under investigation since the U.S. election, according to the press.
Prediction markets also have detractors in the US Senate. In August 2024, eight senators wrote to the CFTC claiming ‘political betting markets interfere with elections and further erode public trust in democracy’ and ‘political bets change the motivations behind each vote, replacing political convictions with financial calculations’. The group supports a proposed rulemaking prohibiting betting on the outcome of the country’s elections.
What are the main risks associated with Polymarket & its peers?
The lack of regulation exposes users to risks such as insider trading. For instance, a market appeared on Polymarket in October 2024, predicting who would be revealed as the founder of Bitcoin at the end of a HBO documentary. It attracted an improbable $44 million, leading to concerns that members of the production team had placed bets.
Markets are also vulnerable to manipulation, such as wash trading, where traders buy and sell a single asset at the same time to boost trading volume and make it seem more appealing. Crypto risk management firm Chaos Labs estimated that wash trading could account for up to a third of Polymarket’s trading during the 2024 US presidential campaign.
Despite the best efforts of AMMs, a lack of liquidity can prevent users from betting on certain markets. While trading volume soared ahead of the election, it fell by 85% after the result was confirmed. This plunge demonstrates how prediction markets rely on major events to drive activity.
Disputes can also delay payouts. Polymarket warned users that they may not receive their winnings until the presidential inauguration in January 2025 (over two months after voting closed) in case the loser contested the result (as Trump did in 2020).
Finally, if oracles fail, trust will evaporate quickly. This scenario hasn’t affected a prediction market yet, but exploits happen. In November 2020, decentralised finance (DeFi) app Compound liquidated nearly $90 million worth of loans after a malicious party manipulated DAI’s price on the Coinbase exchange, Compound’s sole data source.
Current landscape of decentralized prediction markets
At just under $500 million (as of February 2025), prediction markets have a relatively small market capitalisation compared with more established sectors such as gaming ($15 billion) and decentralised physical infrastructure networks ($20 billion).
The biggest player by some distance is Gnosis, with a market cap of $463 million (Polymarket hasn’t issued a native token yet). Originally launched as a prediction market, Gnosis evolved into a platform built on Ethereum to host markets like Omen (still privately owned) and infrastructure provider Azuro. The platform is controlled by a decentralised autonomous organisation called the GnosisDAO and its native token (GNO) allows users to participate in its governance.
Still, Polymarket dominates total value locked (TVL), a metric that measures user deposits in DeFi apps, accumulating over $100 million on its platform. The overall TVL for the sector is $135 million (as of February 2025), ranking it just outside the top 30 use cases according to crypto data aggregator DefiLlama.
Conclusion
Decentralised prediction markets allow users to earn rewards for accurately forecasting the outcome of an event like a political election. By leveraging blockchain technology, they enhance trust and fairness compared to centralised markets and reduce costs.
The betting terms are programmed onto smart contracts, and networks known as oracles gather real-world data to confirm the outcome. Most markets use AMMs to ensure sufficient liquidity for each event.
The arena where prediction markets have made the greatest impact is politics, with Polymarket offering a more accurate forecast of the outcome of the 2024 US presidential election than pollsters. But concerns remain about their potentially negative influence on elections.
There are also risks associated with trading on prediction markets, partly because they’re unregulated, meaning users are exposed to insider activity or market manipulation. Other risks include insufficient liquidity, oracle failures and lengthy dispute resolution.
Augur was the first decentralised prediction market to launch, but Polymarket is the most popular by trading volume. Gnosis, an Ethereum platform that hosts markets, is the biggest project in the sector by market cap.
