As firms look beyond latency to improve trading and risk strategies, on-demand agility is coming into the picture, offering the ability to model and test strategies in accelerated time throughout the trading day. Last week’s A-Team Group Intelligent Trading Summit in New York considered the issues of on-demand agility and how it can be achieved in a panel discussion led by A-Team Group conference chair Pete Harris. Harris was joined by experts from IBM, Kaazing, Datawatch and Xenomorph, who provided examples of their work across the industry and discussed how technology platforms need to evolve to support agile trading.
Brian Sentance, CEO at Xenomorph, highlighted growing interest in different platform interfaces, such as Python, and described how firms are trying to balance agility and latency when looking at solutions like cloud use to perform calculations on a very elastic basis. Craig Kravetz, value engineering at Datawatch, said he was seeing the breakdown of data silos and noted that firms are finding it increasingly important to decide where to position technology. The aim of this is to make technology infrastructure available to connect applications directly to a single source of data.
In terms of bolstering the agility of trading systems, Frank Greco, director of technology at Kaazing, said many firms are pushing the adoption of mobile technologies. He was also quick to stress, however, that the fundamental difficulty with many of these is that the web and HTTP format is not suitable for underlying market data.
With technology evolving at an increasingly rapid rate, Roman Chwyl, world wide financial services & insurance sales leader at IBM Platform Computing, nodded in the direction of the open source community, noting that open source technology is evolving on a daily basis, making it important to sustain an agile platform that can embrace new developments.
Turning to the issue of implementing agile solutions, the panel agreed that computing power is rarely a problem given the speed of machines these days. Instead, the primary problem that lies ahead is the integration of a proliferation of devices. Greco said ‘the iPad changed everything’ and expects iOS and Android devices, PCs and even Wi-Fi to be available in cars in a few years time, requiring demand for mobile applications rather than traders chained to their desks.
Moving on, Chwyl considered what kind of components are needed in a software defined environment. With four main types of infrastructure in the market – high performance, compute, enterprise and application – he mentioned the importance of placing different workload patterns on the correct infrastructure for maximum performance. Kravetz agreed with this assessment, saying the priority is to make more revenue for clients and give them more information on where their money is going. When firms are dealing with large quantities of data, he said the aim should be to deliver only the information that is relevant to specific clients. This means telling ‘the exact right story, to the exact right client given their portfolio’.
Harris then turned the attention of the panel towards evaluating whether there is a need to look across the enterprise and use what is already there. Sentance gave some good examples, mentioning Microsoft’s Excel Power BI, no-sequel databases and the broad variety of social sentiment data that allows people to conduct analysis and deploy it across the business. Kravetz added that the confluence of static and real-time data requires real-time reports to assess how things are going and where adjustments can be made.
Before closing the session, the panel addressed an audience question about how to archive and store large volumes of real-time, millisecond data. Sentance mentioned Rainstor and its scalable technology for archiving, while Harris suggested Sony’s latest 170 terabyte tape that was introduced to handle these types of problems. Chwyl referred to an IBM development with elastic storage aimed at boosting cost performance and supporting user customisation. The panel concluded with a quick question about the future of HTTP, with Greco noting that while HTTP is perfectly adequate for document distribution, WebSocket is the way to go for market data over the web.