Opening Cross: Low Latency Isn’t Just About Low Latency

In last week’s column, I discussed the importance of low-latency data, whether or not you’re active in low-latency trading. And this week’s news makes it hard for me to resist returning to the topic again.

Now, I know some of our readers who deal in areas of pricing and data less impacted by latency sometimes feel like the redundant third guest on a date between the financial markets and latency (Awkward!). But faster is generally better—assuming speed doesn’t impact quality—and if you could do something better, why wouldn’t you?

This week’s front-page stories all give a hefty nod to the importance of speed: UK-based FCM Kyte Group’s use of ITRS’ Feed Latency Monitor solution speaks for itself—the importance of not just capturing low-latency data but also of checking source against source to ensure optimal delivery in reality—as does Tibco’s new FTL messaging platform, which boasts latency of 384 nanoseconds over shared memory. Meanwhile, Thomson Reuters’ use of Connotate’s web-monitoring “agents” to automate an increasing share of its data collection processes isn’t about low latency per se, but is about speeding up data capture and making it more efficient and timely.

Inside this issue, the trend continues: Spread Networks has cut latency on its already-fast Ethernet wave route between New York and Chicago, Greenline Financial Technologies is expanding the message protocols that its systems can monitor, to include low-latency exchange feeds, and in our Open Platform (courtesy of sibling Waters magazine), Charles Ferland of IBM’s Blade Network Technologies subsidiary describes how multi-function switching and routing devices can help optimize low-latency networks.

In addition last week, Mountain View, Calif.-based cPacket Networks publicly released the source code for its time-stamping software that the vendor says adds timestamps to data messages on the wire rather than the network card, eliminating any queuing that could affect latency measurements, while the datacenter space saw a flurry of activity: Datacenter operator FiberMedia announced the opening of two facilities in Westchester County, NY and Secaucus, NJ; while low-latency data vendor Quant-House extended its QuantLink network to Nasdaq OMX’s Sweden datacenter, allowing local clients to access low-latency data from other markets via the vendor; and SIX Telekurs opened two datacenters in Equinix facilities in Singapore to host its local IT infrastructure.

While Telekurs’ initiative may be more about redundancy, reliability and—like QuantHouse’s initiative—ease of market access, hosting in third-party datacenters usually contributes to latency somewhere along the line. So even if latency isn’t the primary objective, it can still be a knock-on benefit.

There are still plenty of areas where latency isn’t a major issue, such as Wall Street Source’s new Research-1 application, which allows users to deliver intra-day news digests to clients or colleagues for small incremental fees—the vast majority of news consumers are in roles that simply don’t need news as it breaks, and can suffice with a regular roundup of the news important to their portfolio, says WSS chief executive John Albert (despite the trend towards low-latency, machine-readable news for algorithmic trading)—or CarryQuote’s IntelliCast platform that enables firms to deliver multimedia research to clients’ mobile devices via a branded display application.

The same goes for indexes, such as the volatility benchmark index launched last week by interdealer broker Tullett Prebon, or the iTraxx index of Latin American sovereign debt launched last week by Markit. However, that doesn’t mean the data should be slow—index values still need to be real time, even if not low latency, and with index providers calculating and distributing index data on a streaming, real-time basis, low latency could yet find uses outside of its current comfort zone as new trading  strategies emerge.