On 20 August 2014, Émile Servan-Schreiber announced that the US version of the Hypermind prediction market had just launched. His short update advertised a €10,000 geopolitical contest, which he said was partly sponsored by the US government. It invited participants to trade and help decision makers see further into the future.
The announcement gives the prize amount in euros, despite describing a US launch. A contest prize is an incentive for participating forecasters; it is separate from the virtual money used to trade on the platform. The government sponsor and the specific contest rules are not identified in the post.
A new market with an older technical history
Hypermind’s official history identifies Lumenogic as the company that launched the market in 2014 to develop an international community of forecasters. A 2015 conference paper by Servan-Schreiber and Pavel Atanasov places the market’s initial launch in May 2014. The August update concerns its US version, rather than establishing the start of the whole market.
The paper describes the early mechanism: participants traded play money, each starting with 100,000 H, while strong performers shared real cash prizes in proportion to virtual profits. It links the underlying technology to Lumenogic’s work for the Good Judgment Project. The paper describes the early platform’s mechanism separately from the specific August geopolitical contest.
The authors also describe invitation-based recruitment from strong performers in earlier forecasting competitions and encouragement to share information. Recruitment selects who contributes; the trading mechanism determines how their judgements enter the market price. The journal’s account of proven forecasters traces the longer history of using performance to identify useful contributors. That distinction helps explain why a market’s design includes both its participants and its trading rules.
Launch evidence and performance evidence
The 2015 paper subsequently evaluated six markets on the 2014 US midterm elections against statistical forecasting models. Its comparison used daily forecasts and Brier scores, with lower scores indicating smaller probability errors. The authors reported an advantage for Hypermind while acknowledging the small number of questions. That is an early evaluation of separate election markets; it is not a result for the advertised €10,000 contest.
The US launch extended a crowd-forecasting market through an offering aimed at an American audience. Its geopolitical challenge invited participants to turn judgement into trades, with a stated prize pool. The later research describes a separate evaluation of election markets. Keeping those purposes apart allows readers to understand the launch and the early mechanism without treating the advertised contest as an established accuracy result.
