Competitive Hedge Fund Management AI and the State of Economics
Come away O human child
To the waters and the wild
With a fairy hand in hand
For the world’s more full of weeping
Than you can understand.
– William Butler Yates
Numer.ai is a hedge fund that is co-managed by tens of thousands of predictive models built by anonymous data scientists, geneticists, statisticians, and physicists alike. These scientists use machine learning models to try and perfect a “meta” model for stock trading. This is done by putting user-submitted learning models into competition with each other over market data from an unknown source. Scientists are incentivized with Numeraire (now trade-able), the site’s own cryptocurrency.
Firstly, why is this legal? Well, little do people know, 84% of trades on Wall Street in 2012 alone were taken care of by neural networks trading on nano-second scales. This puts the average person out of the day trading or margin markets nowadays (at least by pure number manipulation rather than good ol’ research), but if one has a little know-how and a decent computer they easily can design their own algorithm and play along with the delusion. You can also use or buy bots that other people make really easily for cryptocurrency exchanges. Wall street investors have competition to have the closest servers to the exchange just for that extra few nanoseconds of lag time. If a “meta” model is perfected, human trading would be effectively ineffective.
This is all great in the sense that we can just forget about market manipulation and efficiency eventually and go back to a resource-based economy (super-charged by AI resource managers), but not so great in that anybody who’s uninformed is immediately at a massive disadvantage in today’s economy – which is the vast majority of people. Hell, data bots may have elected Donald Trump they’re so effective at predictions, in this case by telling organizers when and where and what information to push. We all really need to keep on our toes with the coming machine learning revolution. This tech will strip some people of their livelihoods and rescue many others from poverty depending on the ethos of the people with the tech, what jobs can be automated, and what the culture transitions to after that (not to mention the impact of climate change). This is re-engineering economies and can rewrite people’s very perceptions if they are managing what content people see.
Do not underestimate these new technologies. Machine learning and parallel computing powers self-driving cars, traffic networks, modern medicine (including bionics), modern research (especially biology), and anything else where statistics and parallelism can be applied (as in all of the rest of science). For math wizards, I recommend learning up on tensor networks. For programmers and engineers, get into Tensorflow or equivalent and learn what convolutional networks, recurrent networks, regression forests, and qubits are. For the average person, do your reading and find a way to take advantage of this tech sooner than later, especially if you’re interested in investing.
Here’s an interesting discussion with Peter Joseph on the reality of the current structure of the economy: