Component architecture is the future of risk management software. Kronos Risk is a pioneering development
With Kronos Risk, London-based Kronos Software has tackled the thorny risk management issues of consistency and performance. In doing so it has used state-of-the-art technology in a precedent-setting way. Moreover, its approach is proving effective in practice.
“Kronos has moved on from the object technology and client-server architecture with which the first generation of risk management systems were built,” says Amir Khwaja, the company’s chief executive officer. Kronos Risk uses the simpler, more pragmatic concept of “components”, which retains the key advantages of self-contained, reusable programs. “It’s easier to build more complex systems with components,” says Khwaja. Among the components in Kronos Risk are user interfaces, calculators and databases, which communicate with each other via middleware.
The
Kronos Risk architecture has multiple, or “n” tiers (where “n”
stands for any number). Kronos keeps the client component as “thin”
as possible – restricting its activities to the access and display
of information. All the work of managing data, pricing instruments and
calculating risk measures is done on servers that can be distributed around
the network for efficient processing.
Kronos has exploited this distributed n-tier component architecture to avoid duplicate functionality. For example, instead of extracting data from a trading system and pricing instruments in a separate calculator, Kronos Risk treats the trading system’s own calculator as a component of the risk management system, eliminating the need to reconcile the results.
The architecture also enables Kronos to achieve near real-time risk analysis. The system responds immediately to events, such as new trades or prices coming in from a market feed. Components “publish” their events, to which other “subscribing” components respond. For example, a position server will respond to a trade and update its value-at-risk.
Instead of creating a separate monolithic system for measuring risk, Kronos uses the component approach to manoeuvre its position servers and user interface into a bank’s infrastructure. It is a flexible architecture that allows users to upgrade their system relatively easily by adding more servers.
Kronos is not the only software developer using component and n-tier technology. But the company – set up at roughly the same time that the technology emerged in the late 1990s – has deployed it to create a new system that is proving its value where it is already live – Copenhagen-based Unibank and Standard Corporate and Merchant Bank (SCMB) in Johannesburg.
The software industry is likely to move to a more component approach, both as a method for single suppliers to build front- to back-office systems and for specialists, say in developing risk engines or Monte Carlo simulators, to provide banks with mix-and-match components to create their trading and risk management infrastructures.
For
many other software firms, 1999 was a year of consolidation. Following
a period of rapid evolution from the mid-1990s, which saw the development
of a range of new systems for VAR, credit risk and enterprise risk management,
the past year has seen many of the major players focused on strengthening
and extending their products without necessarily breaking new ground.
This was the case with market leaders Infinity and MKIRisk, both based
in California, and Paris-based Reuters Risk Management Systems, which
all focused on revamping underlying technology or integrating product
suites. Toronto-based Algorithmics proved an exception, striking out in
a new direction with its WatchDog operational risk management system.
Some suppliers did turn their attention to the needs of other sectors aside from investment banking, particularly investment managers. These firms present a new set of challenges, for instance in the size of their portfolios and the requirement to measure risk against benchmarks rather than in absolute terms.
Although several suppliers have extended their banking-based risk management systems to accommodate some of these requirements, it was not until June that California-based Barra introduced the first system dedicated to the investment community. Guided by specifications set out by leading US firm Putnam Investments, Barra combined its own investment models with the high-performance risk management technology that it gained when it acquired Redpoint Software the previous year.
Much of the challenge of managing risk is in the acquisition and management of data. SAS Institute, based in Cary, North Carolina, has specialised in providing tools for gathering and analysing data for more than 20 years. Following requests from bank customers, SAS has tailored its tools for risk management.
Risk Dimensions, launched in June, makes use of SAS’s ability to access data directly from a wide range of platforms, including all major database systems. It employs the company’s techniques for consolidating data and drilling down through multi-dimensional information to analyse risk from various perspectives. And it offers a sub-system for developing risk models that, among other things, allows users to handle multivariate simulations with non-normal distributions for Monte Carlo analysis.
Perhaps the boldest move of the past year was by California-based Integral Development, which in September reinvented itself as a Web portal – a site that offers a variety of information and services, often from a number of different sources. Following the launch in the spring of a new version of its trading and risk management technology using leading-edge JavaBeans object technology, the XML data standard and a Web-style architecture, Integral transformed itself overnight into a portal, called CFOWeb.com, hosting portfolio and risk management and other financial services.
