The speed at which electronic markets trade requires pricing and execution systems that can perform at sub-second machine rates. New York-based Numerix, having started out as a pricing specialist for more complex instruments, has added intensive computation power to its platform to enable its clients to enter the realm of high-speed market-making, demonstrating a capacity to support structured products across the spectrum.
“A higher degree of performance and automation are needed as you move into market-making,” says Satyam Kancharla, chief product and strategy officer at Numerix.
The company has also made its platform modular in order to offer services that meet the specific needs of its clients. The broadening of services has led to commercial success and deeper penetration within its client base.
“Initially, our usage [of Numerix] was focused on the pricing of hard-to-value exotic swaps linked to IFC’s funding exotics,” says Howard Covert, portfolio officer at Washington, DC-based International Finance Corporation (IFC), a member of the World Bank Group. “Today, IFC uses Numerix more broadly. We are using the next-generation Oneview offering for valuation and for counterparty risk metrics. We are extending its use to structure convertible debt, hybrid capital and structured products.”
In several instances, Numerix has partnered with other firms to extend its functionality. It has worked with Singapore’s DBS Private Bank to develop distribution and the Reserve Bank of New Zealand on lifecycle process support, including limits, risk and liquidity and collateral management. Where necessary, it has also acquired technology to boost performance.
That approach was needed when it sought to increase support for market-making in structured products. Numerix needed a way to run the many calculations that impact the risk and pricing of structured products efficiently, at sub-second speeds. To achieve this, in March this year it acquired TFG Financial Systems, which offers a high-performance, event-driven calculation framework based on ‘graph technology’.
“A dependency graph is a technology for connecting server systems that speak to each other in an event-driven, real-time, efficient way,” says Kancharla.
TFG’s streaming directed acyclic graph technology triggers calculation modules only if the data they depend on changes, instead of recalculating an entire model in response to an event such as a change in price or credit.
A higher degree of performance and automation are needed as you move into market-makingSatyam Kancharla, Numerix
“As a result, the system can calculate Value at Risk and Greeks and carry out stress testing on a sub-second basis, or price 25,000 interest rate swaps per second in volatile markets,” says Kancharla.
By incorporating the TFG technology into Oneview, Numerix has significantly increased the efficiency of its software, reducing the number of cloud processors an institution might need for a task and thereby lowering the cost of operating its system.
It has also evolved the Oneview platform so that different types of business can take advantage of its strengths without being constrained by how it is put together. Numerix serves a broad market base, including banks, hedge funds, insurers, asset managers and pension funds. Each client has its own requirements and way of working, often with differing requirements among its group functions, such as quants, risk or reporting.
To offer more tailored ways of using the platform, Numerix decoupled Oneview’s individual elements from each another and added interfaces and programming tools that allow institutions and groups to access the functionality as a set of customisable ‘business services’. The programming tools are based on the versatile and easy-to-use Python language, which is less reliant on arcane syntax than other programming languages, and the interfaces sit on an open, standardised web services architecture.
“There is continuum of use cases, from a quant who wants to drill down into models and do ‘what-if’ analysis, to a trader who wants to make a quick decision on a deal and move on, all the way to a chief finance officer looking at capital on a dashboard,” says Kancharla.
Oneview’s business services and Python tools now give users the flexibility to tailor a set of functions to their unique needs, all based on a consistent set of underlying analytics and processes.
Clients have been quick to take advantage of the new flexibility. Examples provided to the awards judges of institutions already using the business services and tools of Oneview included: a rates trader running government bond analysis to assess price and risk of trades in a fixed income book; an insurance quant constructing a valuation process; an asset manager doing real-world modelling as opposed to theoretical risk neutral probabilities commonly used in options pricing for liability-driven investment; and a bank constructing indexes.
One member of the judging panel says: “It is evident there is broad market use for the Oneview Enterprise Platform, and the fact that Numerix is able to work with many different client segments across very complex lines of business is impressive.”
The week on Risk.net, December 2–8, 2017Receive this by email