In a nutshell
The National Stock Exchange is one of the top five stock exchanges in the world, measured in terms of number of trades per day. Set up in 1994, it forms the bedrock of capital markets in one of the world’s most energetic economies. This intense market needed a risk analysis engine which would allow super-efficient trading while mitigating risks.
Challenges, not problems
Conventional risk analysis engines force traders to lock up too much of their capital in margin money, making the market inefficient. Allowing them to trade with less margin money and yet tightly controlling their risk exposure needs complex mathematical models, not simple business rules. These models need super-computers to run, and no stock exchange has such super-computers.
Solutions, not ideas
The key was to use a large number of relatively inexpensive computers to deliver super-computer performance, and at the same time, deliver the always-on reliability which every world-class stock exchange has to deliver. Not easy. We did it by designing the PRISM (Parallel Risk Machine) which works on a cluster of computers and makes a super-computer unnecessary.
We needed a low-latency communication layer to tie the cluster together. We evaluated PVM, MPI, and a few others, and finally selected MPI. The cluster communication backbone, and the entire architecture, were designed and implemented by our team.
As each trade happens on the exchange, the PRISM system re-computes the risk profile of the buyer and seller by looking at various market parameters and the individual’s past trades, and decides in real time whether he is getting into too risky a position or not. If yes, PRISM blocks the trader in milliseconds, before he can place another trade. This process is repeated at blistering speed, since the NSE clocks several thousand trades per second. Not for nothing is the NSE considered a leader in trading exchanges.
Now that’s value
PRISM is one of those mission-critical systems which end-users never get to use, but which ensures that the systems we all use daily continue to deliver good results flawlessly. The NSE was able to reach unprecedented levels of trading throughput and risk mitigation through the integration of PRISM, and was the only exchange in the world to use real-time Value-At-Risk analysis techniques to build an efficient marketplace.