Numeraire and the Crowdsourced, Crypto, Machine Learning Revolution

Numeraire and the Crowdsourced, Crypto, Machine Learning Revolution

Quietly, amongst the hype of cryptocurrencies (Bitcoin was the best performing asset this past decade) and blockchain, a new and exciting model has appeared that combines the excitement of cryptocurrencies, blockchain, and crowdsourcing all into one. A company called Numerai has adopted all three of these revolutions into one exciting model that is attempting to “build the world’s last hedge fund.” They, in essence, use “advances in structure-preserving encryption to allow for open participation in the problem of stock market efficiency.” 

We find a few things particularly relevant to the market research industry with this new model that Numerai has pioneered.

First, Numerai has embraced open participation. Most computer scientists recognize the MNIST dataset used to train various image processing systems. The machine learning (ML) community has embraced this dataset as an excellent way to begin learning ML. When the dataset first became available, the ML solutions were not very good at recognizing the handwritten numbers and letters. It wasn’t until fast GPUs, shared knowledge of ML algorithms, and open participation that finally, we had machines that matched even the best human recognition of written letters and digits. Numerai learned and applied this concept early on by making their datasets open to the ML community.

Second, Numerai encrypts sensitive data while still “keeping it useful for machine learning experts.” Generally speaking, financial companies don’t want to share their data or ML solutions (much like the market research community) because the “financial incentive for secrecy is strong.” Not only that but in the research world, GDPR and various other similar privacy protection acts protect our data. Numerai has adopted a Homomorphic encryption scheme allowing people to test and train different ML algorithms “on its encrypted data without first” having to decrypt it.

Finally, they have launched a cryptocurrency, named “Numeraire,” to compensate winning ML algorithm authors. Numeraire is an Ethereum token traded on several different cryptocurrency exchange platforms. It allows Numerai participants to either retain or sell their numeraire earnings on a public exchange platform.

The implications for other verticals are vast, and it’s easy to see how this concept someday could power a significant portion of the market research industry. Beautifully, powerfully, and efficiently, the market research industry generates insights from targeted and qualified participation, compensating participants (almost always in fiat currency). While the ML models used in regression or classifications are generally known and shared within MR, they are not necessarily easy to implement. Essentially, a “Numerai” type revolution could revolutionize the market research profession and industry. Could a company launch a version of Numerai for the market research vertical? If Numerai continues its successful run, we guess that this concept will spread to many different industries and professions, including market research.