What is Numerai?

Earn Crypto through your Data Science Know-How

If you’ve logged onto Shapeshift.io or even a big Crypto Exchange like Bittrex, then you’ve probably seen Numerai mentioned. Even if you’re not familiar with Crypto at all, but you’ve heard of Data Science, then you might find yourself becoming interested in Numerai as well. In the case that you have no Data Science experience, from an investing standpoint, what Numerai is trying to do can be considered as striking for the finance world at large.

From a quick analysis of the Numerai project, including where it’s looking to land in the future, we hope to make this as explicitly clear as possible for you.

What is Numerai?

You may have heard of the people called “quants” that work on Wall Street. They’re the somewhat shadowy figures behind the scenes, working mathematical models beyond the common man’s understanding to beat the market averages. In other words, they’re often the reason that a general investment fund or, a hedge fund, “wins,” time and time again.

Richard Craib, the creator of Numerai, saw this and took it a step further. He ran with the idea that the people who know the stock market better than the rest weren’t even on Wall Street, at the time. In other words, he looked at the Quants as they were at the time that Numerai started and he saw something missing. Why weren’t machine learning professionals taking advantage of the traditional finance trend towards big funds relying on specialized mathematical models?

Also, it’s important to note that this was not too long ago. All of this was already happening in 2015.

Numerai has the simple, yet grand vision of being the aggregated network of all of the best predictive models in the AI space.

In the first year, Numerai reportedly quickly attracted around 7,500 data scientists, who in turn added their algorithms to the Numerai platform. This leads to what we can think of as their sub-vision, which is just as important. Numerai wants to become one of the biggest repositories for AI algorithms. Therefore, if an AI project wants to work collaboratively with other teams or even just share some foundational ideas with them, they won’t have to look any further than this one particular platform.

Moving in the Ideal Direction

Going back to their history, in 2016, to make matters even better, Numerai received a further $6 million in funding from Union Square Ventures, a Venture Capital firm that’s known for its smart investments into companies like Twitter and Indeed as well as into Cryptocurrency and Blockchain firms.

According to Union Square Ventures, the best part of Numerai is that anyone can contribute to the platform without divulging more information about themselves than they want to, which relates back to its existence as a Blockchain network.

As of now, when you stop by their website, the first thing that Numerai advertises is the ability to “earn cryptocurrency in weekly data science tournaments.” It appears that this is the specific incentive that they use to attract data science professionals to their platform, which equals out to payments in Numerai coins.

If this is in fact how it works at a basic level, then based on Numerai’s current price, it’s not so bad of an incentive.

With the price of each Numerai coin consistently appearing to sit above the US minimum wage and teams most likely being rewarded with groups of coins, data scientists could do a lot worse, especially for a side gig.

This quick discussion of Numerai’s history up to the present time is helpful to set the scene, but since we’re all about all things AI, we’ll jump into the tech next.

The Tech

Helping Data Scientists spread their knowledge among each other and to the world at large, especially in the field of Machine Learning, sounds excellent.

With a grand vision like this one, however, it’s logical to wonder: how does Numerai work behind the scenes? Furthermore, how exactly do they use Machine Learning to keep their platform up and running?

Before going into any of the technical specifications, keep in mind that Numerai’s big goal can also be described as “becoming the platform where open participation takes place related to solving the problem of stock market efficiency.”

This actually means that a lot of what the Numerai network offers is training data for Machine Learning Algorithms of all shapes and sizes, especially those which are looking to forecast and predict certain aspects of the stock market.

To further set the scene, since most internal data on how such AI algorithms work is and always has been kept private, a high barrier is effectively put up that keeps data scientists from being able to contribute meaningful data towards improving such algorithms.

From these ideas, Numerai began to think about exactly how data like this could be democratized. What they came to first was that encryption is the logical first step, through a technological framework that is called a “structure preserving encryption scheme.”

According to the Numerai CEO, they use groups of these, which are basically Machine Learning Algorithms that are encrypted to the point that they can learn from data without knowing everything about what it is.

In this way, theoretically, a company’s data is always safe while an AI system can continue to learn how to use it in progressively better ways. Related to this, their CEO is known for advocating for keeping the algorithms and methods that run their system, as secret as possible, for a very specific reason.

Overall, they boil this down to the idea that in today’s world, especially in the financial industry, secrecy equals strength.

Overarching everything is the fact that Numerai has made it clear that a large part of what their network does depends on structure-preserving encryption. All in all, this equals out to a network that caters to as much decentralization and therefore, as much participation by data scientists, around the world as possible.

Looking Forward

This may be confusing at this point, but all of this that we’ve spoken about so far is not even everything that Numerai does, related to AI.

In the second quarter of 2018, Numerai has been moving towards creating what they call an “AWS AMI,” which is apparently like a kit that contains everything that a team needs to do any sort of machine learning project.

In connection with this, they call this stage of their future roadmap, “Compute,” because the Numerai network will have enough computing power to “ping” AIs that are already network nodes in order to get internal predictions. With this, we see a possible connection with Oracles in that AIs that sit on the network and feed predictions to their users play almost the same role.

The major difference between the two would be that Oracles act as continuous data feeds as we’ve mentioned in other pieces, so for example, they can give stock prices to influence a network’s decisions.

AIs that feed predictions, however, can give information that is significantly more involved. One key way in which this could play out is that while Oracles can feedstock prices in real time, AIs could feed specific future levels of those stock prices, related to mathematical models that teach it how to make these predictions over time.

As AIs get better with the more training data that they can take in, it is essential that Numerai continues to promote open data submission on its network over the course of its history.

When Compute draws to a close, Numerai also hopes to have ended up with a new use case for its network which comes down to improving itself and increasing involvement related to what it does. As to the rest of 2018, Numerai aims to go to where Golem AI wants to be and beyond.

If they can truly be a decentralized supercomputing network on which just about anything can be done, then they will surely revolutionize the AI industry and others as well.