Howard Oakley
Despite banks investing vast sums of money in computer-based models to increase their returns, they’re no more accurate than chicken entrails.
By now, we’re getting used to wild swings in financial markets, such as the thousand-point collapse in the Dow Jones Industrial Average in a mere half hour on 7 May. On that occasion, blame seemed to rest with what insiders termed ‘an algorithmic error’ in computerised trading systems. In a more destructive vein, at least some of the blame for the recent recession has been laid on mathematical models employed in financial derivatives. The best-known of these is the Black-Scholes model, which earned Robert Merton and Myron Scholes the Nobel Prize for economics in 1997, and many investment workers their phenomenal bonuses in the 1990s and 2000s, but turned sour in the past couple of years.
Ever since their early days, computers and computer-based models have been dear to the hearts of those who make money from money. The early naïve dream that software could forecast the movement of stock markets has given way to the harsh reality that they’re no more accurate than crystal balls or chicken entrails.
In spite of that, banks and others invest huge sums of money in trying to harness technology to increase returns. Since electronic trading seized the London Stock Market on ‘Big Bang Day’, 27 October 1986, the sky has been the limit. Armed with the right gear and a decent Internet connection, you can now make your first million from a recliner in the sun, or blow your inheritance in a few quick keystrokes.
For much of the time, on both shores of the Atlantic, the systems used to run (or undermine) the world’s financial markets have been PCs. For a brief period in the early 1990s, Steve Jobs wowed Wall Street with his sexy black NeXT computers, and all sorts of exotic hardware has been used to run mathematical models, but by and large purchasing preferences have been conservative. This hasn’t been the case with appliances: from 2002, BlackBerry phones have become one of the stigmata of those working in finance, and Apple’s iPods, iPhones and now iPads have become similarly popular.
One of the secrets of BlackBerry’s success is its Enterprise Server, tailored to endear this mobile phone to corporate IT departments. Although Apple has yet to be quite as integral, its support for the development and deployment of in-house iPhone applications is an unusual strength. Connected to the information resources within a corporate network, iPhones and iPads have ample processing power to be potent financial tools. You and I will never get to see these systems, of course, but perhaps our pension funds will enjoy the fruits of their success.
The snag with these otherwise attractive packages is that, unlike humans, their behaviour is usually predictable, trusting techniques that are deterministic rather than stochastic. Financial institutions have sunk considerable sums into research using less-predictable methods such as genetic algorithms and genetic programming, but few of these have proven fit to release from their labs into the wild of the markets. The upshot is that trading and other financial software tools tend to behave consistently, like a herd of sheep rather than a clowder of cats. When one investor or trader is urged to sell, the chances are that everyone else’s software is recommending the same action.
So instead of reflecting the brilliance of the minds that created them, trading tools have come to drive volatile swings that maintain financial instability. It is perhaps no coincidence that the equation at the heart of the Black-Scholes model is expressed mainly in Greek.
