WallStreet & Technology touches on an interesting topic relating to last weeks near 1,000 point fall in the Dow in late US trading.
Data released by Gomez Inc (see chart below), shows many on-line trading platforms struggled to cope with the resulting surge in investor activity, with some platforms showing a 10 fold increase in the time taken from log-in to trade execution, compared to ‘normal’ market conditions. Such delays can rapidly kill hard won client loyalty for a platform.
Gomez Inc, Website response times for top 15 on-line brokerages on Thursday, May 6
I am sure senior management at many of those platforms are struggling to understand why their platforms were unable to scale sufficiently to support user numbers under conditions of extreme market volatility, without excessive latency and delays being introduced.
This clearly highlights the importance at the design stage of ensuring that the platform has been architected for scalability and performance, even under the most extreme market conditions.
Rigorous automated performance and regression testing should be incorporated into a continues build process testing performance and functionality at every phase. Prior to go-live comprehensive stress testing must be carried out, throughout the entire platform stack, to ensure pricing, trading and all end user GUI functionality is performing to acceptable levels and within agreed latency thresholds. Time spent in final UAT testing, will be worth it’s weight in gold!
At Caplin, we are engaged with a number of banks that are building new multi asset trading platforms. From the early design phase we work with the business and technical teams, to ensure their new platform is architected to deliver the performance, scalability, low latency and full range of functionality required to exceed the projections of the most optimistic of business plans, and perform to agreed and acceptable levels even in the most challenging of market conditions.
Filed under: Single-Dealer Platforms, Technology Trends | Tagged: online trading |
Leave a Reply