In the world of investments, the algorithms have been in the ascendancy for a long time now. By some reports, around 80% of cash-equity trades are now executed by algorithms, independent of human involvement. This is noteworthy for a number of reasons, not least because of potential legal issues of having algorithms allocating a very significant portion of the world’s liquidity. For example:
- Most algorithms tend only to be as good as the humans that create them, which means they retain the capacity for (very expensive) error. For example, in the recent case of B2C2 Ltd v Quoine, the Singapore International Commercial Court (a court that often adopts a similar jurisprudence to the English court) considered the mis-execution of trades at an exchange rate that was approximately 250 times market due to an issue with one party’s algorithmic trading software. It held (among other things) that “where acts of deterministic computer programs are in issue, regard should be had to the state of mind of the programmer…at the time the relevant part of the program was written”. This underlines the point that while replacing humans with machines in equity capital and other markets may have advantages, extinguishing the risk of liability for human error is unlikely to be one of them.
- Recent stock market ‘flash crashes’ have been blamed on the feedback loops generated by algorithmic trading, as sudden movements in share prices prompted algorithms to execute trend-following orders that pushed the markets to extremes in a short space of time. Just as people have made money because of computer programmes violently shunting markets in unpredictable directions, so have people lost it. What do you do if you are on the losing side of algorithm-induced volatility? Are you entitled to some form of redress? Perhaps most significantly, is your investment firm’s terms of business sufficiently watertight to exclude liability from losses caused by failure to adjust for market anomalies caused by technical failures?
- Market manipulation may be as old as markets themselves and new technology often, unfortunately, means new opportunities for the unscrupulous. By way of example, algorithms that scrape data from public sources for trading strategy purposes may be vulnerable to being gamed by fraudulent release of misinformation (computer programs often face problems in discerning between legitimate data and “fake news”). That can manufacture volatility and cause loss. Even something as simple as a social media update can have a profound impact, both for listed companies and investment firms (e.g. see Burford Capital’s allegations of potential illegal market activity centring on a hedge fund’s Twitter feed, as reported in August 2019).
- Many modern algorithms are not programmed with traditional “if/then” logic, but rather are structured as neural networks, adopting deep learning techniques that allows for “self-creation” of trading strategies. There may be interesting questions regarding legal liability for losses in circumstances where the human programmer and his/her employer has little to no insight into how the program’s output (i.e. trading gains or losses) is being created. For example, if an investment firm is not aware of what their algorithm is liable to do in any given situation (which is reasonably likely for any firm relying on a neural network/deep learning technology) and, as an investor, you lose money when the algo makes a bad call, is that negligent and/or a breach of the investment firm’s duties to its clients?
These are just a few of the potential legal issues that may arise in this complex and fast-moving world. If you have any dispute-related queries regarding algorithmic trading (or pertaining to a related field such as high frequency trading), please get in contact.
