Posted by Peter Simpson on February 19, 2018

Coping Efficiently with Stepped Up Supervision of Algo-Trading

The Financial Times has a major article out last week about increased scrutiny of algorithmic trading in the UK. Quoting from the article, “The warning from the Bank of England’s Prudential Regulatory Authority, and the Financial Conduct Authority, covers the development, audit and monitoring of trading algorithms, which generate quotes for securities automatically without human intervention… Concerns among trading authorities have been heightened by incidents including a so-called flash crash in the pound in October 2016. A report from the BoE found the crash was exacerbated by an inexperienced trader lacking expertise in algorithms.”

While this may sound scary, it doesn’t have to be. Tools exist that enable traders and compliance officers to monitor and analyze real-time trading activity – even activity happening in algo-driven trading systems. They can spot outliers, anomalies, and exceptions very quickly and then zoom in to the time line to understand the underlying issues. From there, they can decide whether to take action, and what action to take if necessary, with a complete picture of where things are going wrong.

What should be clear to anyone working in eTrading environments, though, is that attempting to conduct such analysis using traditional numerical displays is foolish. There is simply too much data changing too quickly to be able to identify all the trades and/or algos that may be creating issues. Significant data can easily be hidden in aggregated totals, and trying to look at all the data at once is impossible.

Analysis using advanced visual techniques like treemaps, heatmaps, time series scatter plots, and so on, however, is time-efficient and supports a level of comprehension that makes it possible to make fully informed corrective action immediately.

This is not merely a matter of pasting up visualizations on a dashboard screen. Users must be able to interact with all the data, including historical time series and real-time streams of data coming over message buses and CEP engines (aka “streaming analytics”). They need the ability to zoom in and out on selected time frames, filter in any way needed, and change perspectives with just a few clicks. Without these interactive capabilities, a visualization dashboard is barely of little more use than a traditional spreadsheet.

Megan Butler, director of supervision of wholesale markets at the FCA, stated, “Firms should consider and act on [the report of the damage caused by an inexperienced trader] in the context of good practice for their business.”

If you’re not ready, it’s easy to fix that. Contact us to set up a time to see how some of the largest banks, hedge funds, asset managers, and exchanges are using Panopticon Fast Analytics to address exactly this challenge.

The full FT article is well worth reading – find it here.