

Numeraire (NMR) is an ERC-20 token created by Numerai, a San Francisco-based hedge fund that uses machine learning predictions from a global network of data scientists to inform its stock market trades. NMR is the token that powers this system: data scientists stake NMR on their predictive models, earning rewards for accurate predictions and losing staked tokens (via burns) for poor ones. This creates a direct link between AI-driven financial modeling and token economics. Numerai secured $30 million in Series C funding in November 2025, backed by major university endowments, reaching a $500 million valuation.
Numerai runs a weekly tournament where data scientists worldwide submit encrypted machine learning predictions based on obfuscated financial data provided by Numerai. Participants stake NMR tokens alongside their submissions to signal confidence. Numerai aggregates the best-performing models into a “Meta Model” that informs real-world trading decisions. Scientists whose predictions perform well earn NMR rewards; those whose models underperform have their staked NMR burned. In August 2025, JPMorgan committed $500 million in capacity to scale Numerai’s operations, validating the model at an institutional level.
NMR is one of the few tokens whose value proposition is directly tied to AI performance. Unlike most crypto projects that use “AI” as marketing, Numerai’s entire business model depends on machine learning models producing profitable trading signals. The token’s burn-and-reward mechanism creates real economic consequences for model quality. As Numerai’s hedge fund performs better and attracts more data scientists, demand for NMR increases for staking purposes.
NMR price moves with a combination of crypto-wide sentiment and factors specific to the Numerai ecosystem. Key drivers include: tournament participation and staking demand, hedge fund performance and institutional partnerships, token burns from unsuccessful predictions (reducing supply over time), strategic buybacks by Numerai (such as the $1 million buyback in July 2025), and broader market conditions including BTC/ETH trends. The maximum supply was reduced from 21 million to 11 million tokens in 2019, and ongoing burns continue to shrink the circulating supply.
Numeraire is used primarily by data scientists who participate in the Numerai tournament. The platform solves the problem of overfitting in financial machine learning — models that look good on historical data but fail in live markets. By requiring participants to stake real economic value (NMR) on their predictions, Numerai aligns incentives: only confident, well-tested models get staked heavily. For investors, this creates a hedge fund powered by thousands of independent, competing AI models rather than a single internal team.
NMR circulating supply sits around 6.9–7.4 million tokens (as of March 2026), with a maximum supply of 11 million. Numerai distributes NMR rewards from its treasury to data scientists whose models contribute positively to the Meta Model. The staking mechanism means scientists risk their own capital, and payouts are proportional to both the amount staked and model performance. This has attracted thousands of participants globally, creating a crowdsourced intelligence network that Numerai uses to inform real trading activity.