Cookies on KBL website

To improve our website, we use Google Analytics cookies. These small pieces of data placed in your browser show us some of your activities on our website (such as which pages you’ve visited, etc.) and allow us to measure audience on the website. For more information, please visit our Website Data Protection Policy


01 June 2016

Money-making Machines 

“If we all die, it would keep trading.”

Ben Goertzel, co-founder and Chief Scientist at Hong Kong investment company Aidyia, could not be clearer: his team’s new automated hedge fund – run by artificial intelligence (AI) – can keep on running, come what may.

With its capacity to identify and execute trades, this software learns, adapts and evolves by analyzing vast amounts of data without any human input.

Computer-driven investment is, of course, nothing new. For years, numerous quantitative hedge funds have been developing technology and data management that relies on sophisticated algorithmic formulas.

What distinguishes Aidyia and other newer players is the increasing capacity of AI to learn from past mistakes and refine its own parameters – with little or no human intervention.

To that end, such systems can recognize changes in the market and adjust in ways that previous models could not.

For example, deep learning – one of the more promising machine-learning techniques being explored – involves the use of artificial neural networks to recognize patterns in data, and has sparked a surge of interest in AI.

One of the most notable successes for this type of AI was Google’s AlphaGo algorithm, which recently defeated the world champion of Go, an ancient Chinese game so intricate that the majority of experts said it would take at least a decade before machine could defeat man.

Additionally, deep learning powers image and spoken word identification in the kind of applications behind Apple’s Siri, Google’s image search and Facebook’s facial recognition abilities.

Unsurprisingly, investment in these technologies has accompanied their progress, with companies like San Francisco startup Sentient Technologies being backed by $143 million in funding.

Financial firms’ interest in increasing automation is no secret. While an estimated 9% of all fund managers make most of their trades with help from computer models, according to market research from Peqin, a record 40% of hedge funds launched in 2015 used computer models for the majority of their trades.

These automated systems evaluate data such as market prices, volumes and accounting documents to build predictions. They then “decide” on the subsequent action.

Still under development, the goal is for the next generation of AI investment tech to outperform humans. In theory, its capacity to sift through such immense amounts of data should position this technology to uncover better and more profitable patterns of investment.

In other words, if AI can use deep learning to analyze millions of images in order to pinpoint specific aspects of a photo (like faces), it should be able to use the same principle to identify profitable aspects of a stock.

As with any developing technology, though, apprehension and skepticism over the potential of AI remains intense.

To truly flourish, AI requires clearly definable rules that will not be found in the volatility of markets or human behavior. A common belief among skeptics is that the current AI hype, like in the past, will simply fizzle out.