trading is growing more multi-product and multi-region on a daily basis. The high demand for fiber optic networks by trading firms proves that co-location is not the latency panacea.
Reducing propagation delay, network latency produced by the physical distance data must travel, has been and will continue to be the primary method of reducing latency. TABB Group estimated that the $15 billion spent by financial firms on data centers in 2009 was largely driven by a need to reduce propagation delay. This approach is widely known as co-location, and is a hot button topic for all trading electronic markets.
However, co-location can’t solve everything. Not because space near top matching engines is scarce or those that need to co-locate can’t afford it, but because
Also, despite the convergence of data centers in New Jersey for New York based market centers, US-based futures are still traded in Chicago and London Stock Exchange (LSE) listed securities are in London. One should never say never after the happenings of 2008, but it’s safe to say that the CME, which just announced its new data center in suburban Chicago, the LSE, and dozens of other major world exchanges, will probably never move to New York.
However, trading geographically diverse markets cannot be written off because of the inherent latency. The latency impact for firms trading these strategies can easily reach into the billions. Latencies cannot be reduced to microseconds as in co-location scenarios; however, the potential profits of decreasing well below current levels should be enough motivation for the needed innovation.
Access to these low-latency networks is growing. Managed service providers will offer trading firms not only the connection but also the needed data and exchange connectivity. More traditional telecommunications firms lease bandwidth on their fastest lines, leaving the trading firm to connect up and send data as they see fit. And finally, big firms and those whose strategy relies completely on low latency take the do-it-yourself approach by leasing dark fiber and lighting it using the latest and greatest networking equipment.
Despite the availability of low-latency networks and the knowledge of strategies benefiting from the connectivity, just a few can take a piece of this pie. Only so much bandwidth is available on the lowest latency networks so those with money available snatch up the quickest fiber and hold on to it tightly. Also, even a millisecond or more saved on a fast and short fiber network can be squandered if a firm’s trading engine has unwanted latency or is incapable of sending or receiving data at the same high rate as the fiber.
The shortest path has dropped from roughly 1,000 fiber miles to 900 fiber miles, ultimately settling at 825. Without moving churches, schools, mountains, or full cities, any further decrease in the actual mileage seems unlikely for at least a few decades. That means getting the fast fiber on that fast path is the only way to truly be first. We’ve all been taught that slow and steady win the race—well not this time.
The TABB Group Vision Note Long Distance Latency: Straightest and Fastest Equals Profit is based on conversations with trading firms, optical equipment providers, telecommunications firms and high speed trading solution providers. The study provides a detailed description of how advancements in fiber optic technology and network paths can reduce latency associated with long haul networks. Discussion centers on determining the shortest path for your network, types of fiber optic cable that reduce latency and methods for gaining access to the lowest latency connectivity. The study also provides estimates for North American connectivity spending by financial services firms.