Is “AI Trading” Good Idea?
If there ever was a time when bringing artificial intelligence (AI) and trading together seemed alluring, that time is certainly now: trading has gone digital and AI is apparently experiencing its Golden Age. Plus: who wouldn’t want to automate trading?
But, then again, most people like pizza and almost everybody likes chocolate, but chocolate on pizza tastes funny; my point being: not every combination works as well as you would expect. So, is AI trading possible, or is it just a speculation? If possible – is it a good idea?
First things first: let’s make sure everybody understands what we’re talking about.
When we talk about AI, we usually have machine learning (ML) in mind. And machine learning basically means programming computers in a way which allows them to independently learn from input data, developing predictions about the future without being unambiguously told to do so. In other words, machine learning is when you tell an android to walk from point A to point B and it finds a way to do this without being explicitly programmed how to tackle the obstacles in-between.
“Cryptocurrency”, on the other hand, is an umbrella-term used to describe a subset of digital and alternative currencies, i.e. digital money created with the help of cryptography. The idea behind the technology is devising a way to keep transactions both secure and decentralized. In order to do this, an integral part of the world of cryptocurrencies is a process called mining, during which the so-called bitcoin/altcoin miners exchange computational resources for the digital currency. We’ll get to that later.
What we are interested in this article, however, is neither AI, nor cryptocurrencies per se, but the intersections between the concepts. Are there any? And if there are, which are they and how can we use them?
Because, if you have thought about this, but haven’t found an answer yet, you might want to follow me on a short walk down dee’s “what’s new” lane.
AI-powered Hedge Funds
Together with natural language processing, image recognition is one of the two classical AI problems. It took a while for scientists to reach the low error rates of today. And if we took a more careful look at how they finally managed to do this, it would certainly become evident that their success owes everything to these three contributing factors:
- Modernized hardware
- New ML algorithms
All three should be considered when thinking about introducing AI into hedge funding. However, collaboration deserves special attention both because of the fact that it has been largely ignored until recently but also because it is quite possibly the most important – if not key – factor for success.
According to Numerai – a hedge fund created by South African developer Richard Craib, which allows data scientists to control the capital in the fund by submitting their predictive models, – the traditional market model doesn’t perform well and needs significant improvements:
“The problem here is that self-interested market participants trading with all their might to earn more money are not in alignment over a goal. They are adversaries. They have no incentive to work together, to share knowledge, to share data or share code to improve the market. Finance is anti-collaborative, and that hinders progress big league.”
In February 2017, 12,000 data scientists within Numerai’s ecosystem were issued 1 mln crypto-tokens and were assigned to work on improving the hedge fund’s performance by means of AI. For six months since Numerai’s launch, the financial reward was issued in bitcoins. However, NumeraiÂ decided to switch to their own digital currency “Numeraire” in order to encourage further collaboration between its data scientists.
A Stock Market Without People?
Ben Goertzel, chief scientist at Aidyia, confidently claims that:
“If we all die, it would keep trading.”
It, for Goertzel, is Aidyia’s AI trading system and his confidence seems justified. When someone loses a job, someone else saves money. A gradual AI revolution should result in something similar, as far as the stock market is concerned.
Automating everything is not a new idea and has been the driving force behind many patents and industry developments. Unfortunately, we’re far from automation in many fields. In the case of stock markets, though, the only thing not clear at the present moment is not whether it is possible, but how traders – read: people with $500 hourly rates – are going to tackle the changes in the job market.
Since they seem all but inevitable: AI trading systems predict new market trends better than humans; they even trade better than us. It’s obvious why: they can rely on large databases of historical market information to build predictive models, in addition to being able to actively learn using new data. Altogether, this makes AI systems extremely attractive for many top market players. Giant investment banks and hedge funds have already started making the switch; the new players seem to be “born” AI-oriented.
But, then again, you may want to play some part in the world of digital trading without being a data scientist or managing a hedge fund. There is yet another option: blockchain mining. Strong hardware – mainly GPUs – is everything you will need to get started.
OK, you say, but how is this in any way related to AI? Well, this is where the best part comes. Multinational technology giants such as Google (Tensor Processor Unit) and IBM (Watson) are currently working on the development of super smart AI-powered platforms which should boost the ML performance anywhere between 15 and 30 times.
At the same time, Nvidia and AMD – “PC gaming GPU giants” – are planning to release GPUs intended exclusively for mining digital coins, a good news for miners who want to cut spendings on hardware. The writing is on the wall: brace yourself for a new era of digital trading.
Whether supply chains or business communication is concerned, blockchain technology is becoming increasingly interesting for tech experts. The fact is that blockchain technology is rooted in collaboration and trust, but, when AI comes into the equation, these become even more important factors. Unlike in the traditional business model, it’s not enough to create a competing environment in the world of digital trading. One needs to create conditions where everyone’s success depends on the success of the joint endeavour.