Algorithmic trading in FX is rapidly evolving. Until recently, if you said someone did it, you were effectively using an extremely rude four-letter word. Now, it seems that the term has reverted to the original meaning it had in equities: a sub-set of programme trading designed to gain better execution.
Algos have had a huge impact on equities and it has been widely predicted that their use will grow rapidly in FX. However, because the two markets are very different in many ways, I have at times been extremely sceptical. Perhaps the main differences are that FX is generally a principal business, liquidity is good and that the main liquidity providers do not display full order books (see EBS story below); also the buy side is extremely reluctant to pay for anything. Many are likely to take the view that there is no need to pay for better execution when generally it’s already excellent.
Credit Suisse, one of the leading providers of algo trading solutions in equities, is now pushing its models to its FX clients. The rumour is that it is charging around $75 per million traded. At first glance, this looks extremely expensive and it’s hard to see who want to pay such a price to gain a better execution.
This week, Tabb Group published a report, FX Algorithms: Bringing Best Execution to FX Markets, which Credit Suisse purchased the right to distribute. Laurie Berke, a senior consultant at Tabb, writes: “Less than one-third of FX trading was executed electronically in 2000. Growing to 62% by the end of 2006, it will approach 80% by 2010. With the continued growth in electronic trading and the fragmentation of FX liquidity, FX algorithms will become standard fare on desktops of buy-side and sell-side traders alike.”
Berke continues: “Like equities, finding liquidity in the FX markets will become the name of the game. Innovative sell-side firms and vendors in the equities and derivatives markets are bringing analytics, transaction cost measurement tools, rules-based strategies and algorithms to their traditional asset-management clients. On the buy side, traders will avail themselves of sell-side tools offered to help them find FX liquidity while preserving alpha and lower transactions costs.”
The report makes an important point: “Because not all dealers have not wanted to jeopardize their competitive advantage, some are reluctant to offer all the tools their clients need to improve trading capabilities.”
From what I hear, one of the things that make Credit Suisse’s algo offering so different is that the models are executing on behalf of clients in an intelligent capacity on multiple liquidity pools. The bank is not just taking the other side of the trade. So, while the explicit cost may look expensive, Credit Suisse is not actually in a traditional adversarial relationship with its client. It is in its interest to deliver models that work and get a quantifiably better execution. What constitutes that is a moot point, but the bank’s agnostic approach will find favour with certain segments of the buy side, and it may well be rewarded for daring to be different.
