One of the buzz terms in the European financial markets in 2008 is "low latency". In the past year we have heard announcements from most stock exchanges about upgraded trading platforms, brokers differentiating their direct market access programs on execution speed, and no shortage of "ultra low latency" network providers promoting light-speed performance.
Banks want lower latency services, but few are able to measure their own performance down to the level of granularity to test these claims. Most are talking about eliminating milliseconds, with a cadre of top firms concerned about microseconds. So how low is "low enough" and where will it all end?
In the electronic trading world, performance is relative - as long as one firm has an edge over the market, competitors will strive to improve their positions, but whether this will continue to the nanosecond mark remains to be seen.
In Europe some interesting trends are emerging, with a knock-on impact on IT infrastructure. Firms spent the early 2000s consolidating their trading infrastructures alongside wider IT strategy, but the markets have followed the opposite trend. In a post-Mifid 2008, there are more venues in which to trade stocks, which will probably result in a fragmentation of liquidity. There are also an increasing number of market data sources, which firms must ingest into their algorithmic order management systems.
In today's algo-trading environment, having a consolidated system trading multiple markets simply won't cut it - firms need to be able to react to regional price movements immediately in that (and other) markets.
Some network providers are often requested to provide lower-latency connectivity. Although there are still some improvements that can be made from optimising network performance, if firms are serious about removing latency, they need to tackle the largest latency component - propagation delay, attributed to the speed of light over fibre. The only way to eliminate this is to physically move their servers across geographies to get closer to the exchanges. Rather than moving offices, the best solution tends to be shipping some servers to a secure third-party datacentre located closer to a trading venue.
This trend is not likely to go into reverse, and, if anything, the use of "proximity services" will create a new benchmark for electronic trading performance. Even two years ago the prospect of shaving five milliseconds between a bank and an exchange would not have been compelling - now, owing to advanced trading systems and in-house technology, a five millisecond delay is considered an outage.
The European and Asian markets have some catching up to do for exchanges to offer the ease of proximity access available in the US. Many of the new entrants to the European low-latency market have been US brokers and hedge funds, but the European institutions and exchanges are catching up quickly. ●