It’s 10 a.m. Do you know where your latencies are? In the fast-moving, fragmented marketplace in 2008, speed became an increasingly critical factor in making trading decisions.
Brokers stepped up latency tracking in their network infrastructure, the market data they receive, and the message response rates from exchanges and ECNs. That enables them to rapidly identify and scoop up liquidity wherever it resides, and to discern problems with tardiness in data transmission as they arise.
This year’s need for latency monitoring and measurement boosted a cadre of niche vendors and sharpened the latency-benchmarking focus of others. Corvil, Endace, Trading Metrics, Correlix, CodeStreet, Voltaire, Solace Systems, 29West and other firms saw greater customer demand for low-latency data and infrastructure solutions. Each addresses aspects of the latency arms race, from monitoring data travel times to hunting for latencies in a network to flushing out bottlenecks and delays in the messaging layer of a trading infrastructure.
This year, exchanges and market centers built out their low-latency market data feeds and focused on achieving sub-millisecond response times for high-volume market participants. BATS Exchange’s average order response time, for instance, is now under half a millisecond, according to the exchange. Among major venues, the New York Stock Exchange is the only relative laggard, although that’s expected to be rectified next year when NYSE Euronext unveils its low-latency Universal Trading Platform.
Rishi Nangalia, head of business development at Goldman Sachs Electronic Trading, said at a recent Financial Times conference that statistical arbitrage funds and many high-frequency trading shops this year began operating at the microsecond, rather than millisecond, level. A microsecond is one-thousandth of a millisecond, which is a thousandth of a second.
Low latency is vital for high-frequency automated cash equities and options market makers. “Latency is the factor [automated market makers] look at most, and that affects that group more than any other group,” said Joe Ratterman, president and CEO of BATS. “The longer it takes to modify or cancel and replace an order, the more they’re at risk to the market.”
For these firms, low latency has become a business necessity, not a choice. But the ability to respond instantaneously also affects liquidity provision. David Weisberger, head of global electronic market access at Citi, points out what he calls a “liquidity paradox” in the public markets. “Ideally, a market maker would be first at a price level in thickly traded stocks,” he said. “When the order is executed, it’s preferable [for that market maker] if there are still orders on the book at that price. So that puts a premium on the ability to cancel in microseconds.” If the orders behind the market maker disappeared, he explained, other firms may have caught wind of changes in the market before the market maker and canceled their orders. This “simple form of latency arbitrage” could therefore affect top-of-book liquidity, according to Weisberger.
Co-location ramped up this year for ultra-speed-conscious firms. Co-lo refers to the placement of a firm’s servers near an exchange’s or ECN’s matching engine to shrink latencies associated with data transmission over physical distances. Nasdaq, NYSE Arca, Bats and Direct Edge have seen many more firms co-locating servers in their facilities this year.
However, 2008’s speed obsession may have led some market particpants astray. Citi’s Weisberger points out that what’s most important for institutional investors remains figuring out “where’s the best chance of getting a fill for a difficult order.” For most buyside firms and hedge funds that aren’t in the high-frequency arena, “low latency tends to be marketing bluster,” he said.
(c) 2008 Traders Magazine and SourceMedia, Inc. All Rights Reserved.
http://www.tradersmagazine.com http://www.sourcemedia.com/
