Numeraire (NMR) is a decentralized open-sourced ERC-20 utility token developed by the team of crowdsourced quant hedge fund Numerai. The platform organizes competitions where data scientists create machine learning models based on artificial intelligence for prediction of stock market moves.
The company Numerai appeared in late 2015 in San Francisco, and then a year later a cryptocurrency NMR was released. The founder and CEO of the platform is Richard Craib, who was an asset manager when this business idea came to his mind. He understood that the secret character of data influenced the products they created in this or that way, thus forming a certain financial model. Therefore, he decided to launch a hedge fund distributing data among its users who can produce models with it and make them work conducting diverse investment strategies.
The founders of Numerai understand that the strongest machine intelligence can be helpful in a wide variety of applications. Even the most challenging technical issues such as inefficiency in the stock market become solvable with the enormous potential of these machines. Numerai launched the project to solve these problems.
The reason for the inefficiency of the stock market is the lack of open participation. It may seem that data is more available nowadays to individuals and there are many free data sources. But it’s impossible to build an effective model with publicly available data, while high-quality stock market data is well protected by data monopolies and hedge funds. These entities don’t want to part with something that provides a competitive edge for them. Under such circumstances, only a small group of people have access to valuable data which makes breakthrough achievements and progress towards more efficient markets almost impossible.
These people used to hide the methods, techniques, and data of an efficient financial model that they’ve developed due to the strong financial incentive for secrecy. This doesn’t let them improve their work. But there’s a way to share the data with other machine learning experts using cryptography methods. Numerai platform enables that, providing the possibility to work jointly over the problem, attracting the best minds in the sphere with their unique approach to issues.
What is interesting, the whole project is a confluence of technologies that didn’t even exist some time ago in cryptography and artificial intelligence. Now, the machine learning models, which were trained on data from blockchain-based marketplaces can employ their huge potential to develop the most powerful AIs. The use of private machine learning helps to use sensitive private data without revealing it. As a result, it’s possible to sell the data, keeping it private on the open marketplaces.
The system is powerful thanks to the possibility to attract the best data globally that is coming from thousands of sources, create open competition between models and algorithms providing also transparency in rewards. At this, privacy is preserved at all stages. Secure computation becomes possible to such technologies as homomorphic encryption, zero-knowledge proofs (ZKPs), and secure multi-party computation (MPC) and Truebit protocol for verification.
A special mechanism was developed by the Numeraire team to allocate the prize pool among the participants who submit their bids. It is based on a multi-unit Dutch auction with several new rules added.
How Numeraire works
The platform arranges competitions among scientists on a weekly basis. During these competitions, scientists obtain access to updated data and try to create financial models using encrypted data sets. The results are uploaded to the platform. Then the testing period follows when all the models devised are tested in a real-world environment and 100 best models are selected from them. The winners are rewarded with bitcoins.
The participants should stake a certain amount of NMR for the prevention of tampering. Those who devised successful models will get the staked amount back together with the BTC reward. The scientists who lost, lose their stakes as well, and their NMR tokens are burnt.
Paying the stake, the data scientists express their confidence in the model they offer. There is a staking prize pool in every tournament. The best models submitted by the data scientists are combined into a tradeable “metamodel”. Only those who created highly effective models get payment.
The exchanges trading NMR include such platforms as Upbit, Yobit, Bittrex, etc. As this is a token made on the Ethereum platform, so it’s possible to store NMR on any Ethereum-compatible wallet. As for the hardware wallets, Ledger Nano S and Trezor are recommended as the most reliable ones. The desktop alternatives include Exodus and Mist. Those who need a web wallet should select MyEtherWallet. ETHAdress is the right option for a paper wallet.
The project had no ICO. The distribution of 1 million NMR tokens took place on February 21, 2017. The recipients of tokens were 12,000 community members who were participants of the competition and whose models performed well.
Max supply for NMR equals to 21 million tokens. The founders of the project have no intention to sell tokens to prevent speculators who are not going to use the tokens. After the initial distribution, a certain number of Numerai is minted on a weekly basis until the max supply is reached.
It is possible to earn Numeraire tokens only by staking NMR. The total amount of earnings on the platform surpasses 140,000 USD per month thus turning this enterprise into the most well-paid data science tournament in the world. One can also earn Numerai reporting vulnerabilities, contributing to open-source packages such as Numerox and performing other tasks, for instance, translating technical documents to other languages.
Richard Craib, the founder of Numerai, was named “Forbes 30 under 30” awardee in 2017. The members of the directory board are Georey Bradway and Xander Dunn. The advisors of the platform are also famous figures in the cryptocurrency world such as Coinbase co-founder Fred Ehrsam, Polychain capital founder Olaf Carlson-Wee and Ash Fontana, a board member at Kaggie.