The vote by the British people to exit from the EU was a huge shock to many people, not least the financial markets who had priced the currency markets and the equity stock markets for a ‘remain’ win. On Friday 24^th June, we saw sterling drop by almost 10 per cent against the US dollar.

These huge swings in the markets, and the associated massive trading in the assets put a huge strain on the IT systems of the banks and the trading venues. One ECN (Electronic Communication Network – a forex trading venue) reported volumes of over $24bn in the day, up from a previous high of $17bn earlier in the year. This is a 40 per cent increase in trading volumes over the previous peak volume the ECN had seen.

This level of extreme trading will be seen across other asset classes, including shares and fixed income products, as the ramifications of Brexit are played out in the markets. And the UK hasn’t even actually invoked article 50 of the treaty to initiate the withdrawal from the EU.

These spikes in trading volumes are a serious problem for businesses. The risks are more than just an outage or failure of the systems; slow performance by the IT systems can cause extra latency in your trades. This is potentially more dangerous as you are trading behind the market and can lose money to those who are more performant and able to ‘trade inside you’, knowing that your trades are delayed and able to predict what you will do next.

For the application support teams, understanding the performance and availability of their very complex systems, and understanding the performance and capacity that is left in the system is critical, but very hard to do in real-time as the volumes are climbing. But modern technology is now able to ‘watch’ the IT systems and to spot ‘hot spots’ as they grow, and identify anomalies or issues as they are building.

In situations like this where there are many variables involved and they are all changing at speed, it is not enough to monitor static parameters. You need a real-time view and a dynamic IT solution that can not only process the sheer volumes of data being produced, but do so in real-time, with an ability to respond dynamically to the continuously moving thresholds. Machine-learning algorithms can do so by learning from the trends and patterns produced from the data and evolving accordingly. Accompanied with automation, anomalies can be noted and responded to immediately without the need for human intervention.

For example, a severe increase in the latency on a critical price feed could trigger a temporary suspension of trading on those instruments while the issue is investigated and resolved. Equally, if the CPU on a particular server is regularly exceeding the 90 per cent mark, some of the processes being fed through that server could be temporarily fielded to a different server. In the long term, this methodology can also be used to manage your server estate more efficiently by gaining clear insight of CPU levels and the triggers that have resulted in severe increases, or decreases, in server activity.

It may be some time before the risks of Brexit across various asset classes are fully transparent and understood, but businesses that can be dynamic and flexible will be best placed to weather the storm. The elasticity of full capacity management, backed up by insights and analytics, is a good place to start.

Guy Warren is the CEO of ITRS.