Author: Max Bowie
Source: Inside Market Data | 30 Sep 2011
Categories: Events
Topics: European Financial Information SummitMorgan StanleyBarclays CapitalEuroMTS
The ever-increasing speed demands of algorithmic and high-frequency trading are prompting a shift to more proprietary technology development, leading to a division between the “haves” and “have-nots” with the resources to continue investing in the fastest systems, while other components of the markets struggle to keep pace, according to panelists at last week’s European Financial Information Summit.
Even within the same organization, some firms are having to balance differing needs for speed versus functionality, depending on their user base. “I’m optimizing the latency component of systems depending on who my clients are,” said Raj Sambasivan, vice president at Morgan Stanley. “Some are very much on the cutting edge at the high-frequency level, while there are others who want more functionality.”
And as more firms pile into the high-frequency space, competition on speed alone is becoming harder to achieve. “In high-frequency trading… we are now at a point of diminishing returns. More people are trading, opportunities are getting harder to find… compared to five years ago, when HFT was based on relatively simple technical decisions,” said Mark Holt, head of technology for systematic trading at Bluecrest Capital Management.
However, though the market is becoming more accessible overall, operating at the true cutting edge is becoming more expensive, creating a barrier to entry. “Increased competition and rising costs as we get closer to zero latency means that other key performance indicators such as jitter and reliability come into play,” said Bob Giffords, an independent banking and technology analyst, who moderated the panel.
“We’ve got closer to the raw data coming from exchanges, which has forced us to look at how we deal with data and structures. We used to allow vendors to do this for us, but now we need to understand this ourselves,” as vendors add extra latency-introducing hops to the data path, Holt said. “We want any enrichment taken away, because it stops us understanding the data, which can be more valuable than speed alone.”
At the same time, getting closer to the raw data means firms must remove other steps from the chain, even as they increase their focus on latency monitoring. “We used to monitor latency using log files, but logging is a big cost, so we don’t have that anymore—so we have to look at the packet capture file,” Sambasivan said.
But the more granular monitoring required as markets accelerate further is creating another barrier to entry. “We are having to measure microsecond differences, so we need nanosecond measurements to do that… and standards are struggling to keep up. For example, the Pcap (packet capture) standard only uses microseconds, so we need to use proprietary formats,” said Richard Croucher, European lead for high-frequency engineering at Barclays Capital. “As a result, the whole area becomes a barrier to entry. We’re now seeing increasingly exotic solutions starting to appear that require different skills and technologies to deploy. So I think we will see fewer players in the hardcore low-latency space.”
In fact, latency information can become as voluminous—and as expensive to generate—as market data itself, making it important for marketplaces to strive for standard practices and performance, panelists said. “We do see differences between different exchanges [and multilateral trading facilities]—some are very fast, and others are also fast but limit the amount of data you can send them per-second… but when they do that and clients send more than that amount, it causes microbursts and queuing,” Sambasivan said.
But maintaining an orderly, fast marketplace without incurring additional costs for members is a challenge itself, said Fabrizio Cazzulini, chief technology officer of European bond trading platform MTS. “We have to balance three factors—latency and jitter, capacity and throughput, and cost. So we need a clear roadmap, flexibility, and to make sure we don’t oversize the platform so that we remain cost-efficient,” he said.
More from Inside Market Data
Related Articles
Latest Media
Events
Updating your subscription status
Subscribe to WatersTechnology
WatersTechnology has been designed with our end-users in mind so now you can pick and choose what content you wish to subscribe to and make considerable savings.
Visit our subscribe page now to see which WatersTechnology subscription package suits you.
Events
Training
Email Alerts
Latest Whitepapers
New regulations such as Basel III are changing trading and risk practices by rewarding banks that actively manage their risk exposure at an enterprise...
Supporting multiple securities identifiers imposes an operational burden that adds cost and latency to critical trading processes. Bloomberg’s recently...
Visitor comments Add your comment