The London Stock Exchange turns to Accretive to help prepare for growth.

Challenge
As the exchange set record levels of trading activity month to month, the question became: "Was there a limit to LSE's systems?". Looking at a series of upgrades and system enhancements, the exchange wanted to know if these would result in a system able to comfortably handle the volumes predicted in the business plan.

Accretive was asked the specific question: How do we handle the largest trading peaks while still ensuring fairness? Fairness requires that trades on a security are executed in the sequence received but not during peaks trade volume can be highly concentrated in small number of securities. Also, modern trading strategies depend upon managing positions across a portfolio of securities. If all trades are not executed immeditaly, the risk of the position increases.

The art is to design the system so that bursts of trades in small sets of securities can be executed fast enough to avoid a backlog of trades.

Solution
Accretive built a series of model's of the trading system. These showed that just building a larger system e.g. more processors, more memory would not be effective in raising the system ceiling (the maximum processing rate.) Instead the models identified issues in the design and core algorithms that were acting as constrains on throughput. (Interestingly, if the system was measured under a load that did not match the pattern of order arrivals in the peak, the measure of throughput and response time would be very misleading.) With Accretive's models it was possible to study the dynamic system behavior under loads that had not yet been seen in the market.

Benefits
The insights provided by the GATE^TM model provided the data needed to target tuning and enhancements to the current system and led to changes in the definition of requirement for the next generation system at the LSE.