How Prediction Markets Work
And why they're one of the more honest ways to track geopolitical risk
When people ask how likely it is that China invades Taiwan, the usual answers come from think tanks, government officials, and journalists — all of whom have various incentives to be cautious, alarming, or just plain vague. Prediction markets offer something different: a number, backed by money.
The basic mechanic is simple. You can buy a YES contract on a question like "Will China attempt a naval blockade of Taiwan before 2026?" If you're right, the contract pays $1. If you're wrong, you lose what you paid. The going price of that contract — say 12 cents — is read as a probability: approximately 12%.
Because real money is at stake, people who participate have a genuine incentive to get it right. That's the core insight: markets aggregate information more honestly than any panel of experts where being wrong carries no personal cost.
How accurate are they?
Better than you might expect, at least for political and economic events. Research consistently shows that well-traded prediction markets are well-calibrated — events priced at 70% happen roughly 70% of the time over large samples. They outperform polls on election outcomes and tend to react to breaking news faster than any analyst.
The catch is rare events. Wars, coups, and major crises don't happen often enough to build a solid calibration record. A market might sit at 5% for years and then an invasion happens — that doesn't mean the market was "wrong" in any meaningful statistical sense, but it's cold comfort. For low-probability, high-consequence scenarios, treat the numbers as directional signals rather than precise forecasts.
Liquidity matters too. A market with $500k traded is much harder to push around than one with $2,000. When you see a dramatic probability on a thin market, it might just be one person's bet, not genuine collective wisdom.
The platforms
The biggest prediction market in the world by volume. Uses USDC (a crypto stablecoin). Not regulated, which means it's accessible globally but has no formal investor protections. Its geopolitical markets are the most liquid and tend to set the reference price that others follow.
The first CFTC-regulated prediction exchange in the US. Uses real dollars, which makes it straightforward to use for Americans. Smaller volume than Polymarket on most geopolitical questions, but growing fast.
Free to use, anyone can create a market. The play-money mechanic means prices are more about belief than financial conviction, but Manifold covers a huge range of niche geopolitical questions that real-money platforms won't touch. Worth watching for questions that don't yet have a market elsewhere.
More of a structured forecasting platform than a market. Users submit probability estimates and are tracked on calibration over time. The community includes professional analysts and domain experts. Good for questions with longer time horizons where careful reasoning matters more than speed.
How to read the numbers on this site
- Probability % — the current YES price. 12% means roughly 1-in-8 odds at this moment.
- Volume — total USD traded on the contract. Higher is more trustworthy. Ignore anything under ~$5k.
- Liquidity — capital immediately available to trade against. Low liquidity = prices easier to move.
- Composite score on the dashboard is a simple average across all tracked markets — a rough barometer, nothing more.
Frequently asked questions
What is a prediction market?
A prediction market is a platform where people trade contracts on real-world outcomes. If you think something will happen, you buy a YES contract. If it happens, it pays out — if it doesn't, you lose your stake. The market price (e.g. 30¢) is read as an implied probability (roughly 30% chance).
Are prediction markets accurate?
Studies show prediction markets are generally well-calibrated on political and macroeconomic events — events priced at 70% tend to happen about 70% of the time. They update faster than polls or expert panels. However, they struggle with rare, high-stakes events like wars, where there's little historical data to calibrate against.
What is Polymarket?
Polymarket is the largest prediction market platform, using USDC (a crypto stablecoin) as its currency. It is not regulated and operates offshore, but has the highest liquidity and fastest price discovery for geopolitical events.
What is Kalshi?
Kalshi is a US-regulated prediction market exchange, regulated by the CFTC. It uses real US dollars and covers political, economic, and geopolitical events. Being regulated makes it more accessible to US-based participants.
What is Manifold Markets?
Manifold Markets is a play-money prediction platform. Users trade with 'mana', a virtual currency with no real monetary value. This makes it free to participate, but prices may not reflect the same financial pressure as real-money markets.
What is Metaculus?
Metaculus is a community forecasting platform where users submit probability estimates on questions. It uses reputation points rather than money. It has a long track record and a strong community of careful forecasters.
Can prediction markets predict war?
They can reflect trader sentiment about war risk, but they're not reliable predictors of specific military events. Rare, sudden events are inherently hard to price — markets can stay low right up until something happens. They're best treated as a real-time sentiment gauge, not a forecast you should bet your safety on.
How should I interpret the probability percentages?
A market showing 15% for 'China invades Taiwan in 2025' means traders collectively price that outcome at roughly 1-in-7 odds. Higher volume and liquidity means the number is more reliable. Very low-volume markets (under $10k) can be moved by a single trader and should be treated with scepticism.