Program trading

In program trading, orders are generated by a software program instead of being placed by a trader taking a decision. More recently, it is rather called algorithmic trading. It applies only to organised markets, where transactions do not depend on a negotiation with a given counterparty.

A typical usage of program trading is to generate buy or sell orders on a given stock as soon as its price reaches a given threshold, upwards or downwards. A wave of stop sell orders has been largely incriminated, during the 1987 financial crises, as the main cause of acceleration of the fall in prices. However, program trading has not stopped developing, since then, particularly with the boom of ETFs, mutual funds mimicking a stock-exchange index, and with the growth of structured asset management; an ETF replicating the FTSE 100 index, for instance, sends multiples of 100 buy orders, or of as many sell orders, every day, depending on whether the fund records a net incoming or outgoing subscription flow. Such a combination of orders is also called a basket. Moreover, whenever the weight of any constituent stock in the index changes, for example following an equity capital increase, by the issuer, new basket orders should be generated so that the new portfolio distribution still reflects that of the index. If a program can generate more rapidly than a single trader a huge quantity of orders, it also requires monitoring by a financial engineer, who adapts its program both to the evolution of the market and, now, to requirements of the banking regulator checking that it entails no market manipulation. Some trading rooms may now have as many financial engineers as traders.

The spread of program trading variants, many of which apply similar techniques, leads their designers to seek a competitive advantage by investing in hardware that adds computing capacity or by adapting their software code to multi-threading, so as to ensure their orders reach the central order book before their competitors'. The success of an algorithm therefore measures up to a couple of milliseconds. This type of program trading, also called high-frequency trading, conflicts however with the fairness principle between investors, and some regulators consider forbidding it .

With order executions coming back, the mutual fund's manager as well the investment bank's trader must update their positions. However, the manager does not need to revalue his in real time: as opposed to the trader whose time horizon is the day, the portfolio manager has a medium to long term perspective. Still, the manager needs to check that whatever he sells is available on his custodial account; he also needs a benchmarking functionality, whereby he may track his portfolio performance with that of his benchmark; should it diverge by too much, he would need a mechanism to rebalance it by generating automatically a number of buys and sells so that the portfolio distribution gets back to the benchmark's.

Trading rooms are also used in the sports gambling sector. The term is often used to refer to the liabilities and odds setting departments of bookmakers where liabilities are managed and odds are adjusted. Examples include internet bookmakers based in the Caribbean and also legal bookmaking operations in the United Kingdom such as William Hill, Ladbrokes and Coral which operate trading rooms to manage their risk. The growth of betting exchanges such as Betfair has also led to the emergence of "trading rooms" designed for professional gamblers. (reference: Racing Post newspaper 19/7/07) The first such establishment was opened in Edinburgh in 2003 but later folded. Professional gamblers typically pay a daily "seat" fee of around £30 per day for the use of IT facilities and sports satellite feeds used for betting purposes. Today there are eight such trading rooms across the UK, with two based in London - one in Highgate and one in Canary Wharf.