Prediction markets may feel like a modern invention, but their roots stretch back centuries. Renaissance merchants speculated on papal succession. London coffeehouses in the 1700s traded odds on political outcomes and shipping routes, markets so influential that they eventually helped shape modern insurance. In 19th-century America, newspapers printed election odds next to financial news, and early exchanges like the New York Curb Market routinely saw election betting volumes rival stock trading.
The idea has always been the same: use markets to aggregate dispersed knowledge and express it through prices.
What is new is the ability to scale this idea globally.
Blockchain rails, mobile distribution, real-time data feeds, and a digitally native population have allowed prediction markets to finally reach the scale required to unlock their full potential. And scale is everything: the more participants, the deeper the liquidity, and the more diverse the information, the more accurate prediction markets become.
Today, prediction markets are evolving from historical curiosities into a new layer of global financial and information infrastructure.
The full report dives deeper into Renaissance betting markets, London’s coffeehouse finance, early U.S. election markets, and how these prototypes mirror today’s onchain architectures.
Traditional forecasting tools like polls, pundits, and surveys rely on sentiment or expertise. Prediction markets work differently. They use financial incentives to reward accuracy and penalize errors, forcing participants to reveal what they truly believe. Prices update continuously as new information arrives, transforming scattered insights into a single, real-time probability estimate. This mechanism is why prediction markets often outperform expert models, political polling, and sentiment-derived analysis.
Across thousands of resolved markets, prediction platforms exhibit Brier scores around 0.09, placing them among the most accurate large-scale forecasting systems ever measured. Crucially, these forecasts remain well-calibrated far from resolution, often surfacing shifts in sentiment well before headlines or official data releases.
Modern prediction markets differ in architecture, design, and regulatory approach, but they all revolve around the same four layers: how markets are created, how trading works, how collateral is managed, and how outcomes are resolved.
Modern prediction markets span a wide design spectrum. On market creation, approaches range from Kalshi’s fully permissioned, regulator-approved contracts, through Crypto.com’s controlled listings and Polymarket’s curated model, to fully permissionless systems like MYRIAD, Limitless, and Opinion, where anyone can spin up markets tied to media content, short-interval crypto moves, or macroeconomic releases.
Trading and liquidity architectures are just as diverse. At one end, Kalshi and Polymarket operate order-book markets, with Polymarket blending offchain matching and onchain settlement. Limitless sits in the middle, combining a CLOB with onchain AMMs to support faster, higher-frequency trading, while MYRIAD and Opinion lean fully onchain, using AMMs and shared liquidity pools designed for composability and capital efficiency. Across all models, custody and settlement, whether centralized clearinghouses or onchain escrows, shape how capital is held and how liquidity scales.
Resolution mechanisms complete the spectrum. Kalshi and Crypto.com rely on centralized, predefined data sources, while crypto-native platforms increasingly turn to decentralized oracles. Polymarket uses UMA’s Optimistic Oracle and Chainlink, Limitless anchors outcomes to Pyth price feeds, Opinion deploys a multi-agent AI oracle for macro data, and MYRIAD resolves markets through EigenLayer-secured EigenCloud.
The complete report expands on each platform’s history, unique features and architecture, as well as comparisons, quotes from builders, and a closer look at the emerging design space.
Despite these very different design choices, all platforms display the same unmistakable pattern: acceleration.
Monthly notional volume across major platforms surged from less than $100 million at the start of 2024 to over $13 billion in November 2025, a staggering 130x increase in less than two years; open interest sits at over $500 million. Activity has scaled just as quickly: monthly transactions rose from a few hundred thousand to more than 43 million as more markets were created and traded. User participation shows the same curve, with Polymarket growing from roughly 4,000 monthly users to more than 400,000, and new entrants like Opinion, Limitless, and MYRIAD onboarding tens of thousands of traders within weeks of launch.
Prediction Market Transactions
Source: Dune (@datadashboards)
This activity is matched by predictive performance. Across thousands of resolved markets, prediction markets achieve Brier scores around 0.09, placing them far ahead of polls, expert models, sentiment measures, and even many financial forecasting benchmarks. Accuracy consistently exceeds 90%, and both Polymarket and Kalshi demonstrate strong calibration, with implied probabilities closely matching real-world outcomes (e.g. a 70% market truly reflects an event that happens roughly 70% of the time). Importantly, data shows that accuracy increases with deeper liquidity
Forecasting Accuracy by Method
Source: Keyrock, Brier.fyi
Markets remain most active across sports, politics, and crypto, with Sports driving daily trading flow and politics anchoring large open interest, especially during major election cycles. But growth is no longer sports-led. In 2025, non-sports categories are driving expansion across both volume and open interest: Economics and Tech & Science lead volume growth, while Economics and Social & Culture dominate open interest growth. As a result, prediction markets are broadening from entertainment-driven speculation into tools for macro, policy, and financial insight.
Critically, this momentum isn’t the result of a single platform. It reflects a category-wide shift: more visibility, more liquidity, better forecasting accuracy, and a rapidly expanding user base across the entire ecosystem.
Kalshi and Polymarket, Aggregate Category Open Interest Growth, 2025
Source: Dune (@datadashboard).
The full report goes deeper with dashboards, category-level trends, user cohorts, open-interest trajectories, and comprehensive accuracy/calibration analysis, for a data-rich snapshot of the sector’s explosive growth.
The scale and acceleration of prediction markets point to clear product–market fit. But growth alone does not explain why institutions, corporations, and platforms are increasingly paying attention. The more important question is what these markets can actually be used for.
A core argument of this report is that prediction markets function as a new class of derivative that provides clean exposure to discrete events rather than the noisy proxies traditional finance relies on. While existing markets can hedge rates, currencies, and equities, they cannot directly hedge the events that drive those variables. Prediction markets close that gap, enabling risk to be managed at its source.
Institutions can use event contracts to hedge elections, policy decisions, inflation releases, or geopolitical outcomes. Corporations can run internal prediction markets to forecast product launches, sales cycles, or operational risks, a method that has repeatedly outperformed standard forecasting approaches. Crypto protocols can use prediction markets to price TGEs, estimate future valuations, and hedge token listing outcomes long before a token actually trades.
Unlike options, prediction markets offer fixed payouts, capped risk, and no greeks, margining, or volatility modeling. As liquidity deepens, probabilities increasingly mirror institutional pricing, co-moving with futures, options, and swaps around major macro events.
Fed Decision in December 2025. CME Fedwatch vs. Polymarket
Source: Keyrock, CME Fedwatch, Polymarket
The full report includes case studies of institutional hedging, corporate forecasting failures prediction markets uncovered, pre-TGE valuation structures, and detailed examples of how event risk can be traded directly.
The long-term trajectory of prediction markets depends on four structural dimensions: liquidity, retention, regulation and integration.
Liquidity remains the primary structural challenge for prediction markets. Because event contracts have binary payoffs and no continuous underlying asset to hedge against, market makers must absorb directional and tail risk, which naturally widens spreads in thinner markets and limits participation. To offset this, leading platforms such as Polymarket and Limitless deploy LP reward programs that subsidize liquidity provision and compensate makers for holding inventory through volatile information shocks. As incentives have improved and professional trading firms with experience managing discrete, event-driven risk have entered the space, order book depth has increased and spreads have tightened meaningfully.
Polymarket Monthly LP Rewards
Source: Dune (@datadashboards)
Retention is also strong across leading platforms. Data shows, for example, that Polymarket outperforms the vast majority of crypto applications in month-over-month user return rates. Prediction markets are sticky because they are habit products: people come back because events keep happening.
Regulation took a major step forward in 2024, when a landmark court decision clarified that political event contracts are not gambling but legitimate financial instruments. Combined with Polymarket’s re-entry into the U.S. via QCEX and Kalshi’s fully regulated framework, the sector now has a clearer legal path than ever before.
Integration may be the most transformative force. Onchain, prediction markets are becoming composable primitives, plugging into wallets, bots, terminals, and structured strategies, all drawing from shared liquidity. Builder programs from platforms like Polymarket and Kalshi are accelerating this shift by enabling permissionless interfaces, with Kalshi in particular extending its originally offchain markets onchain through integrations like Jupiter and Phantom. Offchain, live probabilities are increasingly embedded directly into mainstream products across media, finance, sports, and entertainment, from CNN, CNBC, Google Finance, and Yahoo Finance to broadcasts like the UFC and NHL. Together, these inward and outward integration paths point toward prediction markets evolving into shared liquidity layers with many specialized front ends, rather than isolated platforms.
Daily Polymarket Builder Users and Cumulative Volume
Source: Dune (@datadashboards)
Kalshi Daily Users and Notional Volume via Jupiter
Source: Dune (@datadashboards)
The full report offers deeper analysis of liquidity mechanics and constraints, detailed retention curves, regulatory pathways, and a comprehensive view of how integration is expanding the usage surface of prediction markets.
Prediction markets are not gambling products; they are information markets. Their core function is to aggregate dispersed knowledge into a single signal—a price—that often surfaces what matters in society well before headlines or expert commentary. This mirrors the origins of Lloyd’s of London, where informal coffeehouse bets evolved into one of the world’s most sophisticated risk markets. When incentives align around information discovery, what begins casually can scale into critical economic infrastructure.
The same transformation is unfolding today. Event contracts allow individuals, institutions, and corporations to hedge real exposure, from elections and macro releases to product launches and governance decisions, demonstrating utility far beyond speculation. Their simplicity and flexibility also make prediction markets easy to embed across domains, from media and finance to sports, education, and consumer applications.
As these markets continue to mature, we expect an accelerating wave of adoption from institutions, media platforms, fintechs, and consumer applications that recognize prediction markets as powerful engines for hedging, forecasting, and engagement. The pace of integrations is only beginning. The next chapter is not about prediction markets competing with gambling, but about prediction markets becoming a core layer of our economy.
Dune sits at the forefront of this shift, providing the most comprehensive and well-organized datasets on prediction markets. If you’re interested in exploring, analyzing, or building on top of prediction-market data, reach out to access Dune’s dashboards, APIs, and data infrastructure powering this new information layer.
Looking to use Dune for your company?
