Hypermind sells collective intelligence tools to companies, governments, think tanks and media. Its prediction market is free and runs on play money, and its site says it has been original, live, and free since 2000.
Its flagship content is a report on its own accuracy: 1,141 resolved markets and 981,180 trades from May 2014 to July 2026, with a skill score of 45.6% over chance and a calibration error under 1 percentage point at 5% resolution.
Around the report sit a 4-video animated masterclass, a publications page of research papers, an origin story where every chapter ends with a short line, and four products that each do one job: Crowd, Prescience, Storm and The Forecasting Machine.
Hypermind is a forecasting company that has run a prediction market since 2000 and markets it with evidence rather than ads. This breakdown covers who Hypermind is, the origin story as content, the lines that end each chapter, the press the product earned, research as marketing, the accuracy report, calibration in plain words, the skill score, the money question, the masterclass, lines built to be quoted, the product range, the question types, the Storm process, the clients as stories and what any prediction market brand can copy.
Who Hypermind is
Hypermind builds forecasting tools for organisations. Its home page calls them Collective Intelligence Tools for Decision Makers • Since 2000, and its headline reads Forecast better and prioritize faster with collective intelligence.
The About Us page says the company is Passionate about crowd forecasting and collective intelligence. It was Founded in 2000 as NewsFutures Prediction Market, it is based in New York | Paris, and it serves companies and Governments Think Tanks / NGOs Media. The team is led by Emile Servan-Schreiber CEO & Chief Scientist and Maurice Balick President & CTO.
The front door is the market itself. The home page lists a Prediction Market (Free!) and sells it in short lines: Trade the news that matters. Build your track record & get recognized. Join a smart future-focused community. Original, live, and free since 2000.
Its stated purpose is broader: On a mission to make organizations smarter. The About Us page adds We have pushed the state of the art over two decades through R&D collaborations with scientists from leading laboratories.
- 012000Founded as NewsFutures.
- 022018All activities under the Hypermind brand.
- 031,141Resolved markets in the report.
- 04981,180Trades analysed.
- 0545.6%Skill score over chance.
- 06FreePlay money only, no refills.
“Because the world won’t get better until we get smarter, together.”Hypermind, About Us page (read Nov 2026)
The origin story is content
Most companies keep their history to a line. Hypermind tells it as a long page of dated chapters that starts with In the spring of 2000, Maurice and Emile developed the first version of the NewsFutures prediction market, in Annapolis, MD, USA.
Early chapters show the product as it was: This screenshot is from December 14, 2000. Then The U.S. version of NewsFutures launched in the fall of 2001. Then came the brand. In 2014, Lumenogic launches a new flagship prediction market, called Hypermind, to harvest and nurture an international community of champion forecasters. It quickly attains world-class performance.
The later chapters cover the move into research and products. In 2017, Lumenogic teams with Raytheon BBN and others to compete in a new IARPA research program investigating Hybrid machine-crowd forecasting. In 2018 Lumenogic regroups all its activities under the Hypermind brand. The page ends with So that's how Hypermind came to be. Let's write the rest of the story together!
Every chapter ends with a line
The origin story has a trick that keeps people scrolling. Each chapter closes with a short headline in the first person.
After the first press story: We're live! After going abroad: We're international! After the first sponsors: We're in business! After the book: We're famous! After the study: We're proven!
The tone stays light through the later years too: We're crowd-storming! We're more than tech! We're inspired! We're the future! We're awed! We're powering up! We're policy-relevant! We're making forecasting actionable! The last chapter ends with We're the smart money!
Short lines like these are easy to skim and easy to remember. They turn a dense history into something a visitor reads to the end.
- 01We’re live!After the first press story.
- 02We’re international!After the US launch.
- 03We’re news!After live odds in the news.
- 04We’re famous!After the book.
- 05We’re proven!After the 2004 study.
- 06We’re the smart money!After the accuracy report.
The product earned the press
Hypermind’s early press came from what the market did, not from campaigns. The first big name was the New York Times, which published a story on NewsFutures in its July 2001 print edition. With a news partner, NewsFutures pioneered the integration of live prediction market probabilities with mainstream news media. The page marks it We're news!
Business followed. Corporate America soon took notice and started soliciting the collective predictions of NewsFutures champion forecasters about the business issues it cared about. Then James Surowiecki writes about NewsFutures in the first few pages of his brilliantly titled book.
Television came next. In January 2006, CBS features our prediction market in a Sunday Morning show about the wisdom of crowds. In a world premiere, the segment also includes a brief interview of a champion prediction trader. The page calls it Mom, we're on TV! Soon after, James Surowiecki and Emile are invited to the World Economic Forum in Davos to discuss the wisdom of crowds and prediction markets.
An early product also made the news. In 2008, we invent the Idea Pageant, a prediction-based platform that let hundreds of InterContinental Hotels Group techies propose and filter new service ideas. This innovative platform would later evolve into Hypermind's Storm platform for crowd ideation and prioritization. The Storm page repeats the point: Storm's first use case of this innovative approach with InterContinental Hotels Group was documented by the New York Times in 2008.
Research as marketing
The strongest proof Hypermind owns is research. In 2004, a landmark experimental study conducted by NewsFutures with leading researchers from Stanford and Yahoo! concludes that play-money prediction markets can be just as accurate as real-money markets. This opens the way to wider corporate adoption of prediction markets.
Later projects built on it. In 2010, two consultants join NewsFutures to professionalize consulting in the prediction market space. A later research project went further. It will lead to the development prediction polls, an alternative to prediction markets, of which Hypermind's Prescience platform is a powerful instance. It also proves the existence of a general-purpose forecasting elite, aka Superforecasters, which is the basis of Hypermind's Crowd panel of champion forecasters.
The Publications page exists to Learn more about the science of collective intelligence, crowd forecasting and prediction markets. It lists titles such as Prediction Markets: Does Money Matter?, Prediction Markets: Trading Uncertainty for Collective Wisdom, Intelligence Is Collective, Debunking Three Myths About Crowd-Based Forecasting, Distilling the Wisdom of Crowds: Prediction Markets versus Prediction Polls and Using prediction polling to harness collective intelligence for disease forecasting.
There is a book as well: Supercollective Intelligence: How Collective Intelligence and Prediction Markets Makes Companies, Governments, and Everyone Smarter, by Servan-Schreiber, E 2026, with a link to See on Amazon.com Each paper is a page that search engines and AI tools can find and cite.
- 012004Play money vs real money study.
- 02Does Money Matter?A paper on prediction markets.
- 03Intelligence Is CollectiveA paper on the core idea.
- 04Three MythsAbout crowd-based forecasting.
- 05Markets vs PollsDistilling the Wisdom of Crowds.
- 062026The book Supercollective Intelligence.
The accuracy report
The flagship piece is a report on Hypermind’s own track record, by Emile Servan-Schreiber July 28, 2026. It covers 1,141 resolved markets and roughly one million trades over 12 years, from May 2014 to July 2026. The About Us page explains why it was written: Mid 2026, we analyzed the forecasting accuracy of Hypermind's prediction markets over its full trading history, and found that you don't need money to price the future.
The split is 694 (61%) binary markets and 447 (39%) multinomial markets, with 981,180 trades in all. Readers can Download the report in PDF.
The method is written out. The data is a trade-level price file (timestamp, market ID, outcome, price, quantity) and a ground-truth file recording the winning outcome(s) for each resolved market. To keep the result fair, each of the 1,141 markets is given equal total weight of 1.0, spread across its own trades, so that high-volume or long-duration markets cannot dominate the result.
The page also says where the markets come from. Hypermind operates prediction markets at predict.hypermind.com since May 2014. The topics lean on economics (21%), most often with multi-month or multi-year horizons, and There are notably very few sports markets.
- 011,141Resolved markets.
- 02981,180Trades analysed.
- 03694Binary markets, 61%.
- 04447Multinomial markets, 39%.
- 0512 yearsMay 2014 to July 2026.
- 0621%Markets on economics.
Calibration in plain words
The report explains each measure before it shows a number. Calibration measures how reliably the prediction market prices can be considered as event probabilities. The question it asks is Of all trades placed at a given predicted probability, what fraction of those outcomes actually occurred?
Then it gives the answer in a sentence anyone can read: The calibration is visually excellent, with data points closely hugging the diagonal where reality aligns exactly with the predictions.
The numbers follow. In the 1% precision chart the MACE is 2.65pp when all data points are weighted equally, or a tighter 1.45pp when they are weighted by the number of observations in each bin. With bins of 5% probability each, the MACE is 1.06pp unweighted, 0.96pp weighted. The summary puts it as a mean absolute calibration error under 1 percentage point at 5% resolution.
Teaching the measure first means a reader who has never heard of calibration can still judge the result.
Skill over chance
The second measure is the Brier score. The Brier score is a popular measure of the accuracy of probability forecasts. The report warns that Overconfidence is heavily penalized. For markets with several outcomes, the error is summed across them, giving a range of 0 (perfect) to 2 (worst possible).
The headline result is Hypermind performing well ahead of chance overall (skill score 45.6%), with binary markets more accurate (50.2% skill) than multinomial markets (38.4% skill).
There is a simpler version too. The market priced the right answer as the favourite most often, which resulted in the "correct" forecasting of more than 5 out of 6 binary markets and more than 2 out of 3 multinomial markets.
The report also checks an easy objection. In our data set, 76% of the binary markets had yes/no outcomes. Even though a majority (62%) of these markets resolved on the status quo, always picking the status quo would have scored much worse than the baseline Chance forecast. On the live questions, Hypermind scored even better than it did in the full sample of all binary markets. The summary says the question set skews toward genuinely contested rather than safe.
Answering the money question
The report takes on the biggest doubt about play money directly. A popular misconception is that prediction markets generate accurate probabilities because traders put real money on the line. Its answer: Hypermind achieves remarkable accuracy in the absence of real-money wagers.
The rules explain why. Participation is free and trades are made with play money only. Everyone starts with the same endowment of play money but there are no refills, so poor performers are weeded out while good performers prosper and gain influence.
The comparison data is independent. The report says we use data from Brier.fyi, an independent, open-source project not affiliated with any of the platforms. Hypermind’s own score, which is 0.119 (single-term Brier score), is set beside theirs. The verdict is careful: Hypermind's Brier score accuracy is very likely in the same ballpark as its real-money counterparts. Hypermind is essentially at parity on both Brier score and calibration.
On calibration, All three platforms track the diagonal closely, and Hypermind's calibration error is smaller than its own margin of error, which means we can't tell it apart from a perfectly-calibrated forecaster. The bottom line reads Hypermind shows genuine forecasting skill well ahead of Chance in both binary and multinomial markets, tight calibration at both 1% and 5% resolution, and no reliance on status-quo bias to get there.
The essay closes on the point it has argued for years. A popular assumption is that real financial stakes are what make prediction markets trustworthy. It says "putting your money where your mouth is" is not a necessary condition for success, as I have previously argued in this Bloomberg opinion. The summary is one line: A well-designed "play-money" prediction market, run at a fraction of the cost of the largest real-money platforms, can produce probabilities that are just as accurate.
A masterclass in four videos
The Master Class page opens with a question: How can we better predict the future? It answers that Cognitive scientists know that forecasting is the essence of intelligence.
The format is short: In 4 short animated videos, this masterclass will introduce you to the wisdom of crowds. You will also learn how to improve your own forecasting skills with the ABC’s of champion forecasters.
The four videos are Crowd Wisdom Principles, Crowd Forecasting Super Power, How to Think About Probabilities and ABC's of Great Forecasting. All four credit Animation & Voice Camille Larmanou.
The second video is built on a real study. Before and during the Covid-19 pandemic, Hypermind teamed up with the Johns Hopkins Center for Health Security in a large-scale research study, testing several hundred public health experts and other medical professionals. The findings are stated in two lines: An individual expert’s forecasting skills rival those of a dart throwing monkey. But the crowd’s forecasts outperform the best forecasters in the crowd.
Lines built to be quoted
The masterclass sums up each video in short lines. The first: When properly organized, a crowd can be smarter than even its smartest member. Everyone can be right, even when each person is wrong. Larger crowds are smarter. Diversity and expertise are interchangeable and complementary. Collective intelligence = diversity + independence + aggregation.
The third: The future is not written, it is probabilistic. The only valid forecasts are those expressed as probabilities. A forecast is never simply right or wrong. Use the Brier score to measure the accuracy of a forecast.
The fourth starts with Forecasting is a matter of skill, not luck. Then come the habits. Attitude: the best forecasters love contradictory points of view, stay humble, and change their minds when the facts change. Balance: champions approach a problem by combining a macro “outside” view, using base rates, with a micro “inside” view that considers particularities. Break it down: a complex forecast should be divided into a set of simpler and more measurable elements. Commitment: Forecast often, and in small updates. Don’t overreact to the news.
The About Us page writes the same way: We hold this truth to be self-evident that intelligence is collective, and the worst enemy of intelligence isn’t crowds but conformity. We call this supercollective intelligence. Because the world won’t get better until we get smarter, together. Lines this short are easy for people to share and for AI tools to quote.
- 01Larger crowdsAre smarter.
- 02Collective intelligenceDiversity, independence, aggregation.
- 03ProbabilitiesThe only valid forecasts.
- 04Brier scoreMeasures a forecast.
- 05SkillNot luck.
- 06CommitmentForecast often, in small updates.
One product, one job
Each product gets a one-line job on the home page. Crowd: Our international panel of champion forecasters delivers the reliable predictions you need to quantify and monitor the risks in your environment. Prescience: Our crowd-forecasting (prediction polling) platform is designed to aggregate the informed guesswork of your community into reasoned probabilities and explanations. Storm: Our crowd-storming platform is designed to elicit, prioritize, and evaluate the highest-potential ideas of your community.
The Crowd page leads with The collective intelligence of hundreds of champion forecasters. It helps clients Quantify and monitor important risks or opportunities in your environment, and lets them Access a daily feed of dozens of predictions. They can submit your own forecasting questions to the panel of champion forecasters and access its predictions in real time. A third option reads: Add your own forecasters to the prediction market for internal competition or comparison with Hypermind’s champions.
It sells the idea in one question: Why ask for a second opinion when you can aggregate hundreds? Our forecasters are curious, informed, grounded in reality and rewarded solely on the basis of their performance. Our algorithms result from collaborative research with UC Berkeley, MIT, and the University of Pennsylvania on behalf of U.S. intelligence agencies. When AI and statistical models lack the massive amounts of structured data they need to make predictions, that's when crowd forecasting really gets going. It promises An exceptional track record of accuracy since 2014, and Explainable forecasts accessible through online dashboards, API feeds, and PDF reports of the predictions and their curated rationales. Clients can also send Your own questions posed to the champion panel either anonymously or in a branded contest.
The newest product is AI. Hypermind spins out a new company (TFM ventures Ltd) to launch a powerful AI-powered forecasting tool: The Forecasting Machine. The online SaaS platform combines real-time global news scraping with the supercollective intelligence of an ensemble of leading AI models. Its page states the problem: there are far too few trained forecasters for the vast number of important things that should be forecasted to avoid costly mistakes. Today, AI finally enables quality forecasting at scale. It lists strengths such as Scale: Ask as many forecasting questions as you want, an AI that delivers critical insights about your issues faster than human forecasters possibly could. No other humans needs to know which questions you are asking. Our AI answers longterm and short-term question with the same motivation and thoroughness. Our AI is naturally endowed with machine-like ability to remain intently interested in a question day after day, year after year. The call to action is Try it now: TheForecastingMachine.com, and the page says it delivers fast, affordable, and well-reasoned trending forecasts on any topic.
- 01CrowdA panel of champion forecasters.
- 02PrescienceCrowd forecasting under your brand.
- 03StormIdeas ranked by the crowd.
- 04Forecasting MachineAI forecasts at scale.
- 05Prediction marketFree since 2000.
- 06Master ClassFour animated videos.
Questions in many shapes
Prescience is sold as Reasoned probabilities powered by collective intelligence. It helps clients Collect and aggregate the latent forecasting wisdom of your employees on a large scale, and to find its best people: Identify champion forecasters within your organization. Train everyone to forecast better with objective scoring and feedback.
The service is white label: License and deploy our full-featured online crowd-forecasting platform under your own brand. Ask as many questions as you like, as often as you like, to as many forecasters as you can recruit within your community. Deliverables include a Full-featured crowd-forecasting platform, skinned to your brand, AI-powered curation and synthesis of forecast rationales reveals the story behind the quantitative crowd prediction, a Bilingual user interface in English and French, Automatic keyword-based online news retrieval and Forecasting alarms with email notifications.
Prescience lets you formulate your forecasting questions in many different ways. The page lists five: Binary The correct answer can be yes or no. Discrete Winner-Take-All The correct answer can be one of several discrete outcomes. Ordered Winner-Take-All The correct answer can be one of several contiguous intervals or options. Graded The correct answer is a distribution over several options. Continuous The correct answer is a value in a range.
Scoring modes cover short and long horizons: Fixed Ground Truth Ground-truth forecasts are made relative to a fixed event horizon. Rolling Ground Truth Ground-truth forecasts are made relative to a fixed time window that rolls forward daily. Our patent-pending reward scheme incentivizes long-term ground-truth forecasting using a series of small “drip” rewards along the way. One mode builds on MIT’s Bayesian Truth Serum and Surprisingly Popular algorithms. It suits questions with no quick answer, such as In 2100, what will be the average global temperature?
- 01BinaryYes or no.
- 02DiscreteOne of several outcomes.
- 03OrderedOne of several intervals.
- 04GradedA distribution over options.
- 05ContinuousA value in a range.
- 06ModesFixed, rolling, drip, Bayesian.
The Storm process
Storm is sold as Ideation and prioritization powered by collective intelligence. It helps clients Collect and prioritize a large number of ideas quickly and efficiently. The platform is designed for both mobile and desktop usage. Challenges can run for just a few hours, or up to several months.
It runs in three steps. In ideation, Participants are asked to propose solutions to specific problems. In prioritisation, They do this by distributing a small number of tokens over many of proposed ideas. In evaluation, Decision makers rate the most interesting ideas surfaced buy the collective priorities.
The key choice is prediction over preference: Storm does not solicit their personal preferences (via votes or “likes”) but rather their best guesses on the chances of success for each idea, with rewards for correct predictions.
Results come back in four groups: HIGH PRIORITY ideas judged most likely to succeed by a significant consensus. LOW PRIORITY ideas judged least likely to succeed by a significant consensus. CONTROVERSIAL ideas about which significant cohorts disagree. UNINTERESTING ideas that have not attracted enough interest (pro or con) to be classified otherwise. The reports include Bayesian measures of consensus and controversy.
Clients told as stories
Hypermind tells client work as chapters, with dates and outcomes. In early 2019, the Johns Hopkins Center for Health Security partners with Hypermind to crowd-forecast infectious disease outbreaks around the world. Our Prescience platform aggregates the forecasts of hundreds of public health professionals with those of Hypermind champion forecasters.
Hypermind helps found the School of Collective Intelligence at University Mohammed VI Polytechnic in Rabat, Morocco. It is the world's first multidisciplinary graduate program (Masters & PhD) in collective intelligence.
In 2021, Hypermind starts a long-running collaboration with EDF, the French electric utility that operates the country's 56 nuclear reactors. The page counts more than 50 projects powered by our Prescience and Storm platforms.
In 2024, the OECD’s Strategic Foresight Unit contracts Hypermind to forecast AI risks and opportunities at the 2030 horizon. The results, combining crowd forecasts from Hypermind's champions with those of its proprietary AI-powered forecasting machine, appear in a report the Publications page lists as Futures of Global AI Governance: Co-Creating an Approach for Transforming Economies and Societies.
What any prediction market brand can copy
Hypermind markets a free product with proof. It publishes its own accuracy data, shows the method, explains each measure in plain words and answers the hardest question about its model head on.
The lessons are simple. Write the report people will cite. Teach the subject in short videos. Tell the company story in short chapters with one line each. Give every product one job. Then cut the clearest minute of every explainer into clips and post them daily on TikTok, Instagram and YouTube Shorts from accounts built for short form. For a forecasting site where every challenge has its own leaderboard, see our Good Judgment Open breakdown. For the oracle that marks every milestone, up to 100,000 assertions, with a post of its own, see our UMA breakdown. For a prediction market that grew a newsletter from 30,535 to 55,550 readers in two months, see our Polymarket The Oracle breakdown.
We run creator and clipping campaigns for sportsbooks and prediction markets, with every post checked against the ad rules in each state and tracked for views, link clicks and click-through rate. The system in this article, run for you.
Frequently asked questions
Hypermind is a collective intelligence company in New York and Paris. It was founded in 2000 as NewsFutures Prediction Market, and since 2018 all its activities sit under the Hypermind brand. It runs a free play-money prediction market and sells the Crowd, Prescience and Storm platforms and The Forecasting Machine.
Its July 2026 report covers 1,141 resolved markets and 981,180 trades from May 2014 to July 2026. It reports a skill score of 45.6% over chance overall, 50.2% in binary markets and 38.4% in multinomial markets, and a mean absolute calibration error of 0.96 percentage points with 5% bins, weighted.
No. The report says participation is free and trades are made with play money only. Everyone starts with the same endowment and there are no refills, so poor performers are weeded out while good performers gain influence.
Four short animated videos: Crowd Wisdom Principles, Crowd Forecasting Super Power, How to Think About Probabilities and ABC’s of Great Forecasting. They cover the wisdom of crowds, a study with the Johns Hopkins Center for Health Security, the Brier score and the habits of champion forecasters.
Publish your own accuracy data with the method shown, explain calibration in plain words, answer the money question head on, teach forecasting in short videos, tell the company story in short chapters, and give each product one clear job.
- Hypermind, home page, read 19 Nov 2026.
- Hypermind, About Us page, read 19 Nov 2026.
- Hypermind, Master Class page, read 19 Nov 2026.
- Hypermind, Forecasting Accuracy report, read 19 Nov 2026.
- Hypermind, Crowd page, read 19 Nov 2026.
- Hypermind, Forecasting Machine page, read 19 Nov 2026.
- Hypermind, Prescience page, read 19 Nov 2026.
- Hypermind, Storm page, read 19 Nov 2026.
- Hypermind, Publications page, read 19 Nov 2026.