Low latency can be critical issue

What are the issues concerning latency and how does this impact a fund’s trading?

This is an extended version of an article appearing in Technology Round Table: challenges ahead for the future.

The type of strategy used by a fund determines the issues it will need to overcome. While latency is an obstacle for most, best execution also needs to be ensured together with minimising market impact. The issues are further complicated by high frequency trading strategies.

Latency is primarily an issue in algorithmic trading where micro-seconds count, advises Bob Park at FINCAD. “In the over the counter (OTC) markets, latency is less of an issue because the trade cycles are longer and decisions are made, for the most part, by real people. In that environment as long as you are getting a reasonable level of service by today’s standards from your real-time data provider latency shouldn’t be an issue,” he says.

According to Martin Koopman at FMO if a hedge fund is doing high frequency arbitrage style strategies, market making or market taking, then having a low latency trading system is critical.

“However, the vast majority of funds use more traditional long, long/short, macro, event-driven or other strategies and do a smaller number of trades, maybe one to 100 trades a day. These funds need to ensure they are getting best execution and minimising market impact from their orders, especially as the average trade size has dropped considerably and the high frequency traders are on the other side of the market hoping to make a few cents per share per trade,” says Koopman.

“The most cost effective way is usually to use a broker-provided algorithm to send the trade to the market,” he adds.

Harrell Smith at Portware believes latency is important for long/short funds executing pre-defined orders and routing them to various broker destinations but that it is not as critical for a quant fund running a proprietary arbitrage strategy on a collocated server. 

“Even firms with more traditional trading requirements are concerned about latency,” he notes. “The enormous quantities of market data that systems have to absorb can cripple or significantly slow down order management systems (OMSs), broker execution management systems (EMSs) and other legacy trading solutions. Funds are realising they cannot rely on this kind of technology to support their day-to-day operations if high volatility days which are becoming the rule, not the exception – make their trading systems fall over,” concludes Smith. 

The use of high frequency trading and similar techniques, demands a strong focus on reduction of latency, with each reduction resulting in the opportunity of greater returns says John Cant at MPI Europe. He says faster application is not solely dependent on processing technology but also on how effective architecture components are in working with faster processing speeds. “To do this you need to be able to work throughout the technology solution stack,” says Cant.

“Analysis and tuning to achieve this reduction in latency is not a trivial set of tasks and requires advanced engineering skills and complex co-ordination, so it is important to engage the right people to manage this type of technology initiative,” he concludes.

At Odyssey Financial Technologies, Jon May says an element of end-to-end latency that is often overlooked is that of order generation, the time it takes to turn an investment idea or strategy into an order or set of orders and to test the impact against various scenarios and compliance rules before releasing the proposed trades.

Latency and variation of latency will almost always have an impact on arbitrage and high frequency strategies of alpha generation, says Olivier Vonet at Orange Business Services. “Some arbitrage funds are already looking at ways to modify their order routing strategy if they are able to anticipate latency to specific liquidity pools,” he reveals. He adds: “Anticipation of latency is becoming a trading parameter to the same level that underlying volatility, liquidity or other parameters have always been. So that sub millisecond one way latency information has become a critical variable of some hedge fund strategies.”

Assuming a given trading model is hyper-sensitive to moving markets, overall latency reduction is critical, advises James Parker at Transaction Network Services. “A major issue for funds is determining precisely where latency exists,” he says.

“What is often overlooked is the amount of time lost within the actual systems cycle itself between the application system, algorithm and compliance. Many funds ignore these factors and attack the more easily measured ones such as network components, network design and geographical location,” notes Parker.

He believes funds that build or buy application and network measuring tools be able to make better decisions in latency reduction. “The numbers can be minor, measuring in microseconds to more dramatic millisecond-range changes. For direct access trading, funds with all components tuned in concert will outperform competitors,” he says.