How Hedge Funds are Using AI

I've often thought about how AI technology might be used to gain advantage investing in world markets.  Since markets have been largely digital for a number of years, and trades executed via online platforms nothing new, it would seem that adding an AI element to this environment is not only feasible, but probable.  Michael Lewis's informative 2014 book "Flash Boys", where Lewis details the rise of HFT, or High Frequency Trading, further suggests this possibility by explaining how financial companies are increasingly relying on sophisticated computer algorithms to execute their trades.

Today, Bloomberg Markets is featuring a great AI article about Man Group, a UK based hedge fund with $96 billion under management who has been dabbling with AI technology for at least 5 years.  Initially quarantined in a "nuclear bunker" until 2014, its clear from the article that the rising success of the technology to make profitable trading decisions has simultaneously impressed, and terrified personnel at Man Group.

Developers at AHL, Man Groups technological arm, are quick to point out that using AI successfully is not plug and play, and the trading strategies cooked up by the AI sometimes do not work, or are reinventions of commonly known human trading strategies.

AHL engineers utilize machine learning techniques to train their AI programs to develop effective trading strategies based on deep learning data sets.  Using massive processing power, the machines can quickly scour relevant data like historical trading, company press releases, economic indicators etc. to try to connect the dots and identify patterns or trends that can be used when making decisions.  As described by article authors Satariano and Kumar:

Artificial intelligence goes further by enabling the system to adapt based on the information it receives. At Man, engineers set parameters: exposure caps, asset class, volatility, trading costs, etc. Compliance and risk management rules are ingrained into the system’s DNA, preventing it from going rogue or breaking the law as a fast track to profit. These constraints set up the borders within which the machine does its work. The system then seeks out patterns, making connections among the data that humans can’t see. AI makes educated predictions based on what happened in the past, trading when the odds are in its favor. Man has several such systems at work. The fastest trade several times per day; others hold a position for two weeks or more.

Sometimes the strategies deployed by the AI, good or bad, can be figured out by the human developers.  Fascinatingly, sometimes the developers are unable to explain what strategy the AI is using, and why it is so successful.  This raises an unsettling situation for a hedge fund, where it may not be able to explain to an investor how they are either making, or losing money for them.

For this reason, the fund has been careful to deploy the AI very gradually, and not yet allow AI to make high risk decisions with large amounts of investor money.

As for performance, many hedge funds using AI technology have not yet been able to consistently beat the market (S&P500 benchmark), but it may be reasonable to project that as time goes by and AI becomes more sophisticated, it may start to yield greater returns to those funds with the smartest AI.

At the same time, many rival firms will start to adopt similar AI technology to remain competitive in the investor marketplace, until more and more trades will be made by machines, at speeds that only machines are capable of deciding and executing.  Then, by definition, what we think of as the broader 'market' will be one where machines compete against other machines to gain advantage.

Indeed, there is no limit to where your imagination can lead you on this topic.

Read the full article here.