While the growth in electronic trading has stimulated the application of algorithmic trading, it has lead to some abuses of the term as well as the scale and mode of its application, more so perhaps in the context of foreign exchange trading as in any other asset class. In part, this may be explained because the buy-side institutions and proprietary trading groups increasingly demand the ability to develop and customise complex trading strategies in place of off the shelf algorithms. They require the ability to access not just multiple markets, but to trade across asset classes, through a range of applications.

This article covers three themes, which are the key to mainstream adoption of algorithmic trading in FX markets:

1) Reducing market resistance and objections to algorithmic trading? How much resistance there actually is, and the extent to which objections are substantial or emotional, is a matter of debate.

2) Buy-side growth. The buy-side including, according to some observers, corporate treasuries are perceived by many to be the major driver behind the growth of algorithmic trading in financial markets generally and specifically in the foreign exchange market.

3) The technology challenges. It is only through the successful development and deployment of technology software and the facilities which support it that algorithmic trading capabilities may be exploited fully and profitably.

Reducing FX market resistance and objections to algorithmic trading?
FX: a suitable asset class for algorithmic trading? Foreign exchange is a commodity not a financial instrument, unless it is specifically packaged as such.

Foreign exchange is traded outside an exchange environment on an Over the Counter (OTC), basis and is probably the worlds largest OTC market with an unquantified and, perhaps, unquantifiable number of markets. It is traded mainly on a proprietary basis.

Foreign exchange has too many tradable permutations for the application of algorithmic trading rules, unless the objective is to trade a single block of currency to a single value date ? which, incidentally, may well be the case where the currency risk on a financial instrument is concerned. Most money mangers have far more complex requirements. Do these attributes make it unsuitable for widespread exploitation by application of algorithmic trading techniques and technology? Personally, the author does not believe these attributes make fx unsuitable for algorithmic trading.

Banking consolidation
International banking consolidation over the past decade has resulted in about 60% of FX by traded value being concentrated in 10-12 banks. They control liquidity and the price at which liquidity is available.

The roles of voice and electronic trading (algorithmic trading)
Much of the perceived market resistance or objection seems to derive from the debate about the merits of voice versus electronic trading. This is a perennial debate, and (in my view) an increasingly tedious one, about the merits of each. As electronic communications networks (ECNs) have developed, they have undoubtedly drawn away business from the voice market. Yet there will continue to be a role for the voice business in providing market advice, handling negotiated transactions, working complex and sensitive orders of very high value and providing a description of markets which data is not yet able to articulate.

(Fear of) more regulation
The foreign exchange market has escaped relatively unscathed from the regulation, which is applied, to markets in financial instruments. Algorithmic trading rules enable a concept of best execution to be applied to foreign exchange transactions and, as a consequence, may be seen as giving financial regulators the ability to apply best execution to the foreign exchange markets. It will to a very limited extent under the Market in Financial Instruments Directive (MiFID), where foreign exchange forms a component part of a securities transaction.

A passing phase?
Will the FX market become relaxed about the operation of algorithmic trading within the market? It is happening already driven in most part by the buy-side of the business and the hedge fund industry in particular, principally through cross-asset investment and trading where foreign exchange transactions form a component part of transactions. However, it is as yet difficult, for me at least, to envisage the majority of transaction flow being governed by the application of algorithmic trading rules.

Buy-side growth
Scale of adoption
The author has been unable to find any meaningful figures about the scale of volume or value of transactions executed using algorithmic rules in the foreign exchange market. In part, because it is an OTC market. Furthermore, the majority of such transactions are components of transactions involving other asset classes.

There is a common perception that deployment of algorithmic trading capabilities reduces the requirements and cost for expensive trading desk personnel as well as increasing operational efficiency, increasing the transparency and the speed of trade execution. The management of financial institutions where it is deployed senses that it will enable them to gain better insight and control over the art of trading and be better equipped to confront sharp volatility. This may, up to a point, be correct in equities markets but can the same statement be made about the foreign exchange markets?

The costs of purchase of the technology capabilities for algorithmic trading or their in-house development capabilities have to be carefully evaluated. The costs are substantial. It was estimated that over US$320 million was expended globally on providing the technology components of algorithmic trading. Even higher figures are estimated for 2006. Unless the development of algorithmic trading capabilities is confined to the introduction of relatively simple algorithmic capabilities, it is very unlikely to be a one-off investment requiring some periodic minor enhancement. Rather it is likely to be a continuous investment in developing new and increasingly complex algorithmic technology capabilities based on incessant patterns and correlation derived from constant analysis and comparison of new and historic data. To justify such investment, the technology driving the algorithmic trading has to be a unique and direct contributor to the trading revenues, therefore highly proprietary; certainly unlikely to be bought in from a software provider.

Mainstream asset management
Recent research indicated that the end of 2005 conducted almost 30% of US equities trading using algorithmic trading technology. How much is from the hedge fund industry is a matter of conjecture. The volume of foreign exchange executed through algorithmic trading applications is almost certainly substantially smaller in both value and transaction terms, even though reliable figures are not available. The US fund management industry continues to be a more ready adopter of technology than the mainstream, European fund management industry, in part because the fund management industry is larger they have a central trading desk for transaction execution. By contrast, in Europe the majority of fund managers and certainly the smaller fund managers handle execution of their own transactions.

Cross-asset investment and servicing
Undoubtedly, there is an increase in cross-asset investment where foreign exchange transactions are but one component of the transaction. It is in the aspect of foreign exchange trading where the application of algorithmic trading rules to transactions is likely to increase in proportion to th