Network: How a Monte Carlo walk could save the bank: Only powerful computers can handle the latest complex deals, writes Peter Rodgers

IF THE derivatives trading team at Credit Suisse Financial Products (CSFP) were to think of branching out, they would be well-placed to go into civil engineering. The team is led by David Spaughton, a Cambridge-educated engineer who became a financier a decade ago.

Mr Spaughton's 12-strong team of derivatives specialists mostly have doctorates in engineering, physics or maths. They run the complex computer systems at the heart of derivatives trading. In City circles, they are known as rocket scientists.

The price of a basic financial derivative, such as a future or an option, could be worked out in the Seventies on a calculator with four or five multiplications.

By the mid-Eighties, customers were offered more sophisticated ways of hedging their risks or speculating in the markets. Then, calculations requiring about 10,000 multiplications were needed to set a price. The earliest 8086 laptop machines and perhaps 30 lines of Basic software coped with that.

Today, derivatives are far more complicated. For example, the value of a synthetic mortgage-backed derivative, which mimics the behaviour of an investment portfolio of home mortgages, depends not just on the level of the markets on a given day but on where the markets have been every day during the life of the contract.

The simplest way of pricing these 'path-dependent' options is to use a Monte Carlo simulation on a computer - like a game of roulette by machine. That means feeding in anything from 10,000 to a million random walks (randomly generated paths). A typical Monte Carlo simulation with 100,000 random walks requires 36 million calculations to price a one-year deal.

This needs the power of a good modern desktop computer or work station - typically 32 megabytes of Ram and a Unix operating system to run software sold by specialist suppliers or, in the case of most big banks, developed by in-house specialists.

The rising power of desktops is allowing derivatives to be mixed and matched in ever more complicated products. 'Without computers, you could never keep track,' says Ross Salinger, head of information technology at CSFP.

The scale of the number- crunching emerged this year in the annual report of National Westminster Bank, which for the first time published its derivatives volumes. It had pounds 530bn of derivatives deals in force - more than six times the amount it had lent to customers.

With such amounts at stake, small mistakes in monitoring systems can make a big difference to whether a bank stays solvent as markets crash or soar. The potential for big profits or losses is high.

Banks are increasingly concerned about the effectiveness of the 'back-office' systems which process their dealings in tracking the value and riskiness of derivatives portfolios.

Christiane Banziger, founder of CWB Systems Services, a City firm that troubleshoots for companies with systems problems, says the development specialist's job is to dream up products to make more money. However, the IT department then has to tame the monster that has been created.

A new derivative must be tested, using computer models, to make sure it behaves in a wide variety of market conditions. Before a derivative is put on the market it must also be made to fit the bank's settlement systems, which handle payments. The whole derivatives portfolio must be valued, usually at least daily, so a bank knows how much money is at risk and how its exposure changes as bond and currency markets move.

'The pressure on IT and systems in general is increasing all the time,' Mrs Banziger says. The back-office software cannot keep up with the pace of change in the front office, where the business of developing and selling derivatives is done. 'That is a big problem in the City,' she says.

Traditionally, the business side of a firm - the front office that develops and trades derivatives - puts up a specification for a system, and the IT side tries to meet it. But it takes so many months to develop new back-office systems that business methods have changed by the time the software is ready.

Since the IT department and the business side find it hard to communicate at the best of times, the system often meets only 60-70 per cent of requirements, according to consultants.

Rapid development tools, a type of software widely available in the petroleum and defence industries, would help to bring IT departments closer to the business, Mrs Banziger says.