Spreading power of the 'black box'In the uncertain equity markets this year, there has been at least one clear trend - the growth of computerdriven trading.In often flat overall market conditions, many FTSE companies have found themselves subject to huge, sometimes record, daily volumes of their shares being traded. J Sainsbury, Invensys, EMI, Barclays, Reckitt Benckiser, EasyJet and Man Group have all succumbed to bouts of intense trading where the market appears to descend, pack-like, for a couple of days before it moves on again. It is a sign of the increasing role of what is sometimes known as "black-box" trading - computer -driven trading based on algorithms or mathematical models.Although algorithmic or quantitative trading has been around for a number of years, it has become much more widespread in 2004, particularly trading on market trends. It takes a buy or sell order of a defined quantity and places it into the model that has specific goals decided by the trader. The model generates the timing and the size of the order based on the specific goals of the algorithm. The goals tend to be based on a specific benchmark, price or time. For example, the software used by Credit Suisse First Boston uses large amounts of historical data to help predict trading patterns and real-time data to assess current conditions.It can also control price thresholds and limits and control participation rates and time. TowerGroup, the US research group, estimates that algorithmic trading volumes will double to about 27 per cent of US equity activity by the end of 2006.The UK is believed to be witnessing similar growth and the impact is being magnified by a rush to focus on any clear stock trend that emerges in an otherwise choppy market.Traditionally upsurges in trading volumes in a stock indicated corporate news was about to break, such as a takeover or a profit warning.But often now they can be triggered by even modest initial stock movement driven by vague rumours or news. Philip Green's move for Marks and Spencer was one example where news followed an increase in trading volumes, but often the rumours amount to nothing. Many hedge funds and proprietary trading desks at leading investment banks run sophisticated software applications, sometimes known as "sniffers", which screen equities looking for momentum, effectively "sniffing" out trading patterns. "The software doesn't care what the stock is," a trader at a leading broker said of his company's product."It doesn't look at anything else, such as price/ earnings ratios, it can just look at momentum in stocks." There is little room for technical analysis of historical charts, beloved of many."Typically, a programme could look at the return on prices at a particular level, going back through years of data," said a software programmer at a hedge fund. "Technical analysis is seen as black magic compared to this."When a hedge fund or prop trader then submits their interest in a particular stock, it appears on electronic order books, and the sheer volume of the order is soon noticed."Effectively it works as a massive disclosure of information, from which people trade off," said another dealer. The effect can sometimes be seen in sometimes sharp share price rises and falls on the stock market. "People soon start thinking there's more to it than meets the eye," said one dealer, explaining why the interestmoves so rapidly. "It works on the semi-credible stories best." The success of this type of trading this year has been driven essentially by flat equity markets and the lowvolatility that are the bane of a trader's life. All parties therefore need to squeeze the most out of minor price changes.For the investment banks, algorithmic trading also offers the opportunity to lower trading costs. Among the brokers, the market in algorithmic trading in equities is led by Credit Suisse First Boston, but the other bulge-bracket firms, such as Morgan Stanley, Goldman Sachs and Lehman Brothers, all have significant operations and are trying to capture an increasing part of the business.Hedge funds are also highly active. This provides more revenue streams for the investment banks as the hedge funds seek intermediaries to deal. However, the upsurge in algorithmic trading will have repercussions for traditional traders. "The long-term is not about lay-offs, but rather a fundamental altering of the trader's role," said Gavin Little-Gill,an analyst at the investment management research service at TowerGroup. He forecast that there would be an increasing need for traders with mathematical and programming skills. "These types of individuals will be able to interface most effectively with an increasingly electronic and quantitative environment. We're going to see more and more PhDs on the trading desk."

Check Stephen Blyth column in Risk (Age of reason or age of procedure?) on sophistication from a risk management point of view.

"Black Box" systems have been around for a very long time, and can be found particularly with CTAs. Then you have all the stat arb managers and even long-only managers using models left-right and center.My guess is that programmes will be used more and more, simply because sifting through data is so time-consuming, and very often follows patterns; hence, you can build algorthims doing the work for you. However, the problems with algorithms is that they have a half-life, i.e. after a certain amount of time the returns are diminishing and you need to "re-invent" to stay on top.