Celoxica recently extended its hardware accelerated data feed and execution platform to include foreign exchange and fixed income markets. IntelligentTradingTechnology.com caught up with president Lee Staines to find out what’s driving the development.
Q: Celoxica has added support for foreign exchange and fixed income to its market data and trade execution technology. What are the business drivers for this, for both markets? Do you currently have customers using this new market functionality?
A: Latency is important for algorithmic and arbitrage trading in order to take advantage of pricing discrepancies between asset prices on different markets or differences between the price of an asset and its implied price on a futures or derivatives market. Although this has been applied to equities and futures for many years, other major asset classes are also experiencing an increase in algorithmic trading, particularly where real-time market price information is available from multiple sources (such as regulated platforms or the electronic trading venues of inter-dealer brokers), which provide the input for automated, model-triggered strategies.
The FX market has become increasingly electronic with the advent of FX ECNs such as FXall, Hotspot and EBS. The role of electronic trading in the fixed income market is also growing as some debt products become more electronic and exchange traded. For example, a global tier 1 bank launched a fixed income algorithmic trading platform in 2010 to trade pair strategies in fixed income U.S. futures and cash treasuries and we are beginning to see the advent of fixed income MTF’s such as the launch of Vega-Chi which is the first convertible bond MTF in Europe.
While Celoxica’s products have a definite latency advantage, they also have a significantly lower footprint. Since processing is offloaded via the FPGA cards, more server resources are available for value added processing and in some cases client have been able to reduce the number of servers by a factor 10, collapsing their infrastructure down to one multi-core server and removing unnecessary network hops. This is particularly important in co-location environments where the cost of hosting a server can be significant.
And yes, we currently have clients using our accelerated market data and market access products for both fixed income and FX markets.
Q: Specifically, the support Celoxica has added is to its Generalised Market Accelerator and Generalised Exchange Access. What are these, and what has this entailed? What can they now do that they could not before?
A: We have extended our Generalised Market Accelerator (GMAC) and Generalised Exchange Access (GXA) product suite to include FX and fixed income markets with normalised APIs for ease of integration and time to market. These are now available for 10gE as well as 1gE network infrastructure. Our products comprise hardware, FPGA firmware and software, which are typically embedded in our clients trading servers. As part of our subscription service we provide regular firmware updates, enabling clients to keep pace with the exchanges who make regular changes to their message data formats. Since the FPGA is in effect re-programmable with new firmware, we are also able to keep our clients up to date with new features and enhancements such as enhanced filtering and multiple DMA queue support.
Q: Has Celoxica added and specific data feed or market access support, or is this in the works? Or do you expect others to do this?
A: In addition to supporting the major U.S. and European Equities & Derivatives markets, we have extended the coverage of our market data and market access products to include FX markets and fixed income markets.
Q: What’s your view on the latency requirements of these new markets, versus the equities and futures space where your experience is telling you they are very latency sensitive?
A: Taking FX as an example, we have some clients are interested in our FX container for a single market, where their trading strategy requires the lowest possible latency to get market data to their algorithm on the input side and the lowest possible latency for executions on the output side. We also have other clients who need access to multiple sources of FX market data and execute across multiple venues. In this case, they use a multi-market container. In both cases they gain benefits in terms of latency, footprint (reduction in the number of servers), throughput and deterministic performance (consistency through micro-bursts).
We are also seeing an increasing trend where clients subscribe to our service to replace legacy components so they can focus their internal resources on differentiating their offering rather than continue to build and maintain connectivity to market data and execution venues.