Exchanges can improve clarity and customer service by sharing independent latency data

In the past, traders evaluated exchanges based on their fill rates and prices. But today most orders are cancelled and only a small percentage are filled. According to Celent, high frequency trading now constitutes about 42% of daily U.S. equities share volume and is expected to exceed 50% by mid-2010. A report by New York-based agency broker Rosenblatt Securities says up to 40% of European equity trading is high frequency. Meanwhile data from UBS shows that Asia lags at about 10%. To these players, time is as important as price. To be efficient, traders have to understand price manipulation, changes and trends quickly.
 

Before reaching a trading venue, an order may touch several systems, including those at a Direct Market Access broker, service connectivity provider and co-location provider. Exchanges’ customers want to be able to measure latency between their systems and each of these touch-points. The longer it takes to understand a latency bottleneck, the more likely it is that customers will miss opportunities in the market.
 

Tight integration between execution and investment strategies is critical to delivering risk-adjusted return on investment, or alpha. With explicit transaction costs so low in trading, it is the implicit costs that can destroy the return created in the investment process. Firms must optimize their trading environment to ensure that real return is not lost through market impact and delays. Portfolio managers and traders must be in synch with the urgency of trades and scheduling, select the appropriate algorithms and adjust them as conditions change.

There are several approaches to measuring latency. Some firms may simply look at the statistics provided by their brokers. Others may take a careful look at the latency figures of the networks upon which they depend. Savvier firms measure the entire latency cycle, from market data delivery from exchanges or aggregators, to internal application latency, to order book matching latency. What matters most is that an accurate measurement is taken which weighs all relevant factors and that adjustments are made to lower latency and improve execution. 
 

“In the past when you had a latency issue, customers would immediately stop trading with that exchange,” says Shawn Melamed, CEO of Correlix. “That’s bad for exchanges because they’re not getting the liquidity, and it’s bad for traders because their strategy is down. Every minute the strategy is down, they’re losing money.”
 

The ability to monitor latency helps traders tune their strategy, select the right venues, react to problems quickly and avoid bottlenecks. The more information an exchange can provide its customers, the more likely it is they can create smarter, more sophisticated strategies. That creates stickiness in the relationship.
 

“By sharing the measurements with our participants, we can understand latency across the full transaction chain,” says Mats Andersson, Chief Technology Officer, Global Software Development at NASDAQ OMX. “They know exactly how long it takes to get from their order management system to the exchange’s matching engine and back – for every business transaction, and also how much latency each component in the chain adds.”
 

Many market participants are more likely to trust latency measurements from an independent third party. Correlix is one company that provides a solution for measuring the latency from the customer’s order management system to the exchange or third party connectivity providers such as co-location facilities. Then it measures the latency that the exchange trading system adds for handling a transaction and providing a response.
 

Traders receive the latency data through a user interface and as a raw data stream that goes straight to their black box, just like market data. This allows them to monitor latency changes and tune their strategy accordingly. The solution uses a consistent methodology, so they can more accurately compare the performance of multiple exchanges.
 

“Large high frequency trading shops have many machines running hundreds of sessions to the same exchange,” says Melamed. “They need the ability to immediately understand which session is faster and identify the fastest route to the market.
 

Real-time latency monitoring enables exchanges to pinpoint the cause of a problem, down to the specific component, and fix it quickly. Real-time streaming data enables them to make automatic changes in the infrastructure so the system is self-healing. 

Photo: ISTOCKPHOTO