Can AI-based trading give hedge funds an edge?
Many trading businesses in the hedge fund industry are engaged in automated trading. Several hedge funds (HFs) are considering using technologies leveraging Artificial Intelligence (AI) for the next phase of their trading operation evolution. Some HFs already use AI-based trading.
AI, ML, ANN and EC
The term AI was coined in 1955 to describe the area of computer science that studies machines that work like humans. There are no proper applications yet. AI remains more than a decade away and is still at the stage of early innovation, according to a report by Gartner last year.
However there are companies, including hedge funds and other trading businesses, already using ‘approaches’ to achieve it. Such as Machine Learning (ML) and Artificial Neural Networks (ANN).
Newer HFs use Evolutionary Computing (EC). They employ a trading system which creates random ‘digital traders’ and test their performance against historical data covering multiple market regime changes. Then, the system places the winners in a real market to trade. While using their key attributes to create the next generation of digital traders.
AI relies on data: the bigger the better. In fact, its proponents say that AI is the way to monetise big data. However, HFs eyeing AI-based trading to gain an edge face big challenges too.
AI-based trading challenges
Beside the technology challenges of achieving true AI, HFs and others employing AI-based trading will have to overcome communication challenges. As ML, ANN and EC become more advanced and complex, explaining to clients why trading systems made a wrong decision and they lost money will become increasing difficult.
Furthermore, national financial regulators worry about the combination of financial trading and AI. They want to prevent AI based trading from getting out of control, causing major problems in capital markets. Such as a flash crash or something worse.
In the UK, The Financial Conduct Authority (FCA) earlier this year published Algorithmic Trading Compliance in Wholesale Markets. This is a report helping them to comply with a tough new European Union directive, MiFID 2.
The FCA wants humans to be able to intervene in trading operations, if an algo goes wrong. It also wants senior managers to fully understand the AI technology used. If an algo goes wrong, senior managers will be held accountable.
So far, HFs using fully automated AI-based trading have managed to beat the broader HF industry, but not the stock market, according to a Bloomberg article published at the end of last year. However, AI strategies make sense as a way of diversifying an investment portfolio and the latest Eurekahedge AI Hedge Fund Index does not look bad.
Ultimately, the future of AI in the industry will depend on its ability to generate consistent good returns.