Opening Cross: How Do You Measure Up?

Inscribed over an archway at the University of Chicago—hometown of last week’s Inside Market Data Chicago event—is Lord Kelvin’s quote “To measure is to know. If you cannot measure it, you cannot improve it.” And in an industry so focused on improvement—whether it be improving latency, cost-effectiveness or profits—measurement has become a new obsession.

However, measurement isn’t a new concept to data professionals. In one way or another, the entire data industry is about measurement—measuring supply and demand, measuring the price people are willing to pay, measuring how much something will cost now versus in the future—and on the administration side, measuring how much data each trader, portfolio manager or analyst consumes, how much value they obtain from it, how much they should pay for it, and how they can obtain more value while staying within budget, which itself involves measuring which projects can be completed with the resources available, and the cost-benefit of each, as well as the cost of missed opportunities.

With cost measurement such a priority, initiatives like Interactive Brokers’ new low-cost data terminal, designed to replace more expensive workstations for equities-only consumers (see page 1), or web services data from Xignite—which last week announced $10 million in new funds to expand its data coverage (see page 8)—are likely to attract interest.

Firms measure how each type of content or asset class performs, how one index compares to another—using classification systems such as FTSE’s Industry Classification Benchmark, which the vendor recently licensed to the Chicago-based CRSP (Center for Research in Security Prices)—and how much can be made from trading them, as well as how fast it can be done. And how do we measure speed? Not in seconds, or milliseconds, but microseconds, using tools that themselves require sophisticated time measurement techniques to ensure accuracy at infinitesimally-small time increments, such as those provided by Corvil and FSMLabs (see pages 5 and 8, respectively).

These tools allow the industry to be more accurate when measuring factors such as the physical distance between co-location centers, and even the length of cables across the floor of datacenters (see the write-ups of our Chicago event, starting on page 9), and the number of hops—whether they be a server, a cable, or a separate circuit board or processor—that data must pass through on its way through feed handlers and market connections to reach a firm’s trading algorithm. For examples of initiatives to optimize latency while handling peak data traffic using the latest technologies, see the hardware-inspired efforts of Goldman Sachs’ Sigma X dark pool and Redline Trading Solutions (see page 1) and updates from French vendor Novasparks and US managed connectivity startup Burstream (see page 6).

The industry also measures the capacity required to process the volumes of data generated by constantly-accelerating markets, and order bandwidth and servers to handle the highest peaks. And when we see particularly volatile markets like last month, of course, we measure them to plan effectively for future market events, crunching and re-crunching numbers to guarantee enough headroom, examining historical traffic trends and forecasting future rates—something that should soon become easier as the Financial Information Forum rolls out an on-demand query tool to allow members to download historical volumes data themselves (see page 1).

In short, the concept of measurement is not new to market data; it has just taken the next steps in its evolution, and is now being applied in new ways and to new areas. And if you think something can’t be measured, think again, because chances are the next steps will depend on it, and your superiors will expect you to deliver—after all, that’s how they’ll measure you.