How prediction markets win
If they win.
Prediction Markets: Not About the Money
Prediction markets look like casinos at first glance. You put money down, you win or lose. But the real story is not about gambling or hedging — we already have futures, options, and swaps for that. Those instruments are deep, liquid, and regulatory-blessed.
Prediction markets, by comparison, have historically struggled with liquidity and faced legal scrutiny. As hedging tools, they lose.
Where they can win is as information systems.
Markets as Truth Engines
Prediction markets aggregate dispersed beliefs into a single number: the price. That price is not just odds; it’s a living probability estimate backed by skin in the game.
Robin Hanson, who pioneered much of this field, formalized market scoring rules and argued for markets as truth-seeking mechanisms that elicit and aggregate forecasts. Markets don’t just reflect known data — they incentivize the discovery of unknown data.
Polymarket: The Proof of Concept
Polymarket is a prominent live example. Its markets on elections, geopolitics, and culture have often tracked outcomes faster or more tightly than punditry and many polling averages; major press covered how prediction markets called key 2024 dynamics when poll models lagged.
The magic here isn’t that people make money. It’s that the market becomes an informational baseline — a source of truth in real time.
Information Systems: Why They Matter
An information system is a structured way of collecting, processing, and distributing knowledge to guide decisions.
Traditional systems:
Polls → snapshot of opinions, subject to bias
News outlets → filtered narratives, often slow or politicized
Social media → raw, noisy, unstructured
Expert forecasts → precise but bottlenecked by small groups
AI models → scalable, but risk hallucination and poor calibration
Prediction markets add something new:
Skin in the game → financial incentives for accuracy
Real-time updating → probabilities shift continuously with new data
Distributed aggregation → no need for coordination, just participation
Single actionable number → a live probability anyone can read
This makes PMs less like news feeds and more like live probability terminals — systems that translate chaos into a price signal.
Problems Prediction Markets Could Solve
Where does this information actually help us?
Politics: replacing or augmenting polls with real-time, incentive-aligned forecasts of elections, referenda, or policy outcomes
Finance: predicting central bank moves, inflation trends, mergers, or regulatory rulings faster than analysts
Science: guiding funding and investment by betting on which research is reproducible or which drug trials will succeed
Governance: governments hedging knowledge risk — pandemics, geopolitical conflicts, climate shocks — through probabilities rather than guesswork
Corporate strategy: firms using internal PMs to predict sales, launch dates, or supply-chain disruptions more accurately than top-down projections
Everyday decision-making: a Bloomberg-style PM terminal guiding entrepreneurs, investors, or even citizens in weighing uncertain futures
The Devil’s Advocate
Liquidity: The Structural Weakness
Single Polymarket markets often top out in the tens to low hundreds of millions, while platform-wide volume reached multi-billions during the 2024 U.S. election cycle. By contrast, U.S. options markets clear ~57 million contracts/day and hundreds of billions up toward ~$1T notional/day; 0DTE options alone averaged about $760B/day in 2024.
Thin PMs mean:
Small bets can distort prices
Odds are volatile without being informative
Institutions won’t trust or use the data
Without liquidity, the “truth machine” risks becoming a “toy calculator.”
Regulation: The Legal Ceiling
Intrade was charged by the CFTC in 2012 and later enjoined. PredictIt’s no-action relief was rescinded in 2022, but the Fifth Circuit granted an injunction in 2023 that allowed continued operation. Polymarket paid a $1.4M civil penalty in 2022 and restricted U.S. access.
Update: In Sept 2025, Reuters reported that the CFTC cleared Polymarket’s U.S. return via a CFTC-regulated entity (QCX), materially broadening potential adoption.
Incentives: Truth vs Influence
Large, motivated flows can shape perception, not just reflect truth. Political markets, especially when thin, can be nudged by whales to signal momentum — raising self-fulfilling concerns.
Resolution: When Truth is Fuzzy
Not every question has a clean outcome. “Will X reduce inflation by 2026?” can be debated endlessly, unlike “Who won the Super Bowl?” Ambiguous contracts introduce disputes, erode confidence, and limit utility.
Competition: Experts and Machines
Prediction markets aren’t the only forecasting tools:
Superforecasters (Tetlock’s Good Judgment Project) often outperform other methods
AI systems can process massive datasets and generate forecasts instantly
PMs might not replace these; they may only complement them.
Adoption: Who Actually Uses PMs?
Today’s user base is mostly retail speculators and crypto enthusiasts. For PMs to matter institutionally, they need:
APIs plugged into trading terminals
Dashboards for policy think tanks
Integrations with research ecosystems
Toward a “Bloomberg for Prediction”
The vision isn’t just more markets. It’s systems built on top:
A “prediction terminal” aggregating PM data, expert input, and AI summaries
Real-time dashboards of probabilities for politics, finance, science
Alerts when odds shift meaningfully, integrated into trading desks, policy planning, or corporate strategy
Polymarket is the engine. The terminal is the win-condition.
How PMs Win (If They Win)
Prediction markets don’t beat futures in hedging or beat casinos in gambling. They win if they become infrastructure for truth.
But truth alone isn’t enough. They must overcome liquidity, regulation, manipulation, ambiguous resolutions, competition from experts/AI, and adoption hurdles. They must prove their information is not just different but indispensable.
The optimistic case: PMs evolve into the Bloomberg of uncertainty — a layer of real-time probabilities that guide decisions across society.
The skeptical case: They remain niche, fun, and sometimes accurate, but structurally sidelined.
Both outcomes are possible. That’s the bet.
References
1. Hull, J. (2017). Options, Futures, and Other Derivatives (10th ed.).
2. Hanson, R. (2002). Logarithmic Market Scoring Rules for Information Aggregation.
3. Hanson, R. (2003). Combinatorial Information Market Design.
4. The Wall Street Journal (2024). How the “Trump Whale” and Prediction Markets Beat the Pollsters in 2024.
5. Laudon, K.C., & Laudon, J.P. (2020). Management Information Systems: Managing the Digital Firm (16th ed.).
6. Dreber, A., et al. (2015). Using prediction markets to estimate the reproducibility of scientific research. PNAS.