Futurist resources > Prediction markets
Prediction markets and crowdsourced futures
Last updated July 2026
When I wrote Getting Results From Crowds in 2011, I gave prediction markets a chapter of their own, as both a structure for surfacing collective insight and a tool for anticipating the future. In the years since they have crossed from fringe experiment to mainstream instrument, now trading tens of billions of dollars a year and quoted on financial news alongside share prices.
A prediction market lets people buy and sell contracts on whether an event will happen. The contract pays a dollar if it does and nothing if it does not, so its price is a probability. At sixty cents, the crowd, weighted by the money it is willing to risk, puts the odds at roughly sixty percent. The market aggregates thousands of separate judgments into a single number that updates minute by minute.
Prediction markets have proven themselves as a forecasting tool. Well-traded markets are repeatedly better calibrated than polls or expert panels, because they draw on information held across many people, update the instant it changes, and are disciplined by the money their participants put at risk.
The current landscape is shown below by type of platform, usage, and value.
Real-money prediction markets
These platforms let people trade real money on the outcome of real events, and they carry the deepest liquidity in the field. Polymarket and Kalshi dominate, both now operating in the US with regulatory approval, and the probabilities they generate are increasingly treated as financial data in their own right. The New York Stock Exchange’s parent company has invested close to two billion dollars in Polymarket for that reason.
These are trading platforms, and for many users gambling platforms. Read the prices as signal, not gospel, and keep the incentives of the people setting them in mind.
| Product | Organization | Description |
|---|---|---|
| Polymarket | Polymarket | The largest on-chain prediction market, settling in USDC stablecoin. Deep markets on politics, economics, sport, crypto, and culture. Returned to US traders in December 2025 under CFTC oversight. |
| Kalshi | Kalshi | The first federally regulated event-contract exchange in the US, settling in dollars. Strong on economic indicators, elections, weather, and a fast-growing sports category. |
| PredictIt | Victoria University of Wellington | A long-running political prediction market for US and international politics. Cleared a multi-year legal fight with the CFTC in 2025 and has since lifted its trader and position limits. |
| Iowa Electronic Markets | University of Iowa | The pioneer, running small-stakes academic markets on elections and economic outcomes since 1988. Still operating as a research and teaching tool. |
| Election Prediction Markets | UBC Sauder School of Business | Canada’s academic election markets, reactivated for each major national vote. A useful non-US reference point. |
Community forecasting platforms
These platforms let forecasters predict for reputation rather than money. Each person submits probabilities, the platform aggregates them, and people build track records scored for accuracy over time. This is where careful, long-range, scientific and geopolitical forecasting happens, and for genuine futures work it is the more valuable end of the field.
| Product | Organization | Description |
|---|---|---|
| Metaculus | Metaculus | A community forecasting platform aggregating thousands of forecasters into a calibrated community prediction. The strongest public source for long-horizon science, technology, and AI questions. |
| Manifold Markets | Manifold | A play-money market where anyone can pose a question. In early 2026 it wound down its cash-prize experiment and returned to a pure play-money model focused on forecasting and discussion. |
| RAND Forecasting Initiative (INFER) | RAND Corporation | Crowdsourced geopolitical and technology forecasting to support public decision-makers, continuing the IARPA-funded forecasting lineage. |
| Good Judgment Open | Good Judgment | The free public tournament where anyone can test their forecasting and where new Superforecaster talent is identified. |
| Fatebook | Sage Future | A lightweight tool for recording and tracking your own and your team’s predictions, popular with those building a personal calibration habit. |
Superforecasting and professional forecasting
This strand of forecasting is built on a curated elite rather than an open crowd. Philip Tetlock’s research established that a small group of well-selected, well-trained forecasters can outpredict even professional analysts with access to classified material. Good Judgment Inc now offers that capability to organizations that need rigorous forward estimates.
| Product | Organization | Description |
|---|---|---|
| Good Judgment Inc | Good Judgment | Commercial superforecasting for governments, institutions, and business, including custom forecasts, monitoring tools, and training grounded in the research behind Superforecasting. |
Long-term and accountable predictions
Long Bets is a public registry of long-horizon predictions rather than a market. Named people stake their reputation and a charitable wager on claims that may not resolve for decades, which makes it a rare discipline on anyone tempted to forecast boldly and never be held to it.
| Product | Organization | Description |
|---|---|---|
| Long Bets | The Long Now Foundation | A public arena for accountable, long-term predictions, where forecasters record decade-scale bets and the reasoning behind them. |
Enterprise and internal forecasting
These are internal markets that surface the knowledge scattered across an organization, on questions such as whether a product will ship on time. The notable shift here is away from markets themselves. Cultivate Labs, after fifteen years, dropped them in favor of simpler probability surveys, arguing that good questions and genuine engagement matter more than the mechanism, and that markets can intimidate participants and invite manipulation.
| Product | Organization | Description |
|---|---|---|
| Cultivate Labs | Cultivate Labs | Enterprise crowdsourced forecasting, now built around opinion pools and AI-assisted forecasting rather than internal prediction markets. |
A note on the earlier generation. Much of what this page once listed, including Augur, Gnosis, Predictious, and Inkling, has faded as fees, thin liquidity, and regulation caught up with it. Augur is now attempting a reboot.
How to use them as a foresight tool
Used well, these platforms are a genuine addition to the futurist’s toolkit. Used carelessly, they invite false confidence. A few principles keep them honest:
- Read the price as a probability, not a prophecy. A market at eighty percent is telling you the event is likely, and also that it fails one time in five.
- Weight liquidity heavily. A deep market with millions in play carries real information. A thin one is closer to noise dressed up as a number.
- Triangulate. The evidence is consistent that combining market prices with polls and with community forecasts beats any single source. No one instrument is the answer.
- Match the tool to the horizon. Use the money markets for near-term, resolvable questions, and turn to Metaculus and Good Judgment for the long-range, structural ones.
Three questions are worth asking of any market before you trust it:
- Does this question resolve on a defined date, or is it open-ended? Markets work for the first and fail for the second.
- Is the market deep enough that its price reflects considered money rather than a handful of trades?
- What could move this price that the crowd cannot yet see? That gap is where your own judgment earns its keep.
What they reveal, and what they do not
The case for prediction markets is real. In well-traded, short-horizon questions they are generally well-calibrated, they update continuously as new information arrives, and in 2024 they read the presidential race more steadily than the polls. Combine them with other signals and you have one of the better near-term forecasting instruments available.
The honest limits are just as real. Thin markets are unreliable, and studies of the 2024 races found accuracy varying widely from one contest to the next, with the worst errors where liquidity was lowest. Identical contracts traded at different prices across platforms, and on most days the Harris and Trump prices failed even to sum to a dollar, which is not what an efficient market is supposed to do. A single large trader, the so-called French whale who wagered tens of millions on Trump, could move a price and generate a wave of self-reinforcing coverage regardless of what was actually true. Markets also carry a persistent favorite-longshot bias, overpricing the unlikely and underpricing the near-certain. On genuinely long-range or novel questions, the ones that matter most to the future, capital gets tied up, the informed traders thin out, and the market stops telling you anything useful.
None of this makes prediction markets worthless. Combined with other signals they are one of the better near-term forecasting instruments available. They were simply never built to see the long-range, structural change that shapes the decades ahead.
The frontier: humans, markets, and AI
The most future-facing development in this field is no longer about which crowd is wisest. It is about whether machines can forecast as well as trained people. AI systems now compete directly in forecasting tournaments on Metaculus, which in early 2026 launched a program to track their accuracy against real resolved outcomes. Manifold has declared its ambition to become the best place for AI forecasting and discussion. The early evidence is that the strongest models are closing on elite human forecasters without yet clearly surpassing them.
This is the convergence worth watching. Crowds, money markets, and AI forecasters are being measured head to head for the first time, and the interesting result will not be any single winner but the combination. Humans + AI, forecasting together, will almost certainly read the near future better than either does alone.
Use these markets for what they do well, and read them with a clear eye. For the future that matters most, the one that unfolds over decades and resolves on no settlement date, no market will do the thinking for you. That work is still yours.
