“It only takes four seconds to invest thousands of million of pounds”, said Jérôme Kerviel, the now-imprisoned rogue trader. While it makes me wonder whether I’m in the right business, or on the right network, his and other similar stories do emphasize the very thin red line between huge success and ruinous behavior in the financial trading world, where ultra-low latency financial networks operating at the speed of light create huge repercussions from even a single error — intentional or not.
A Japanese trader, who wanted to sell one share for 650,000 yen, but got the key-in sequence wrong, offered 650,000 shares for 1 yen apiece! The Tokyo Stock Exchange had to close down for several hours to unwind all the millions of purchase placements that followed. Similarly, a Lehman Brothers dealer in London 12 years ago wiped £30bn off the FTSE when he inadvertently ordered sales of shares in blue-chip companies such as BP and AstraZeneca that were 100 times larger than intended.
The reaction to these and similar events has been more network controls and stricter regulation, not only in the forensics department — finding out what actually happened after the catastrophic event — but also better proactive capabilities to spot and stop a disastrous deal from closing at all.
Millions of high-volume trading deals are being transacted every minute across global financial networks like BT’s Radianz and Orange Business Services Flexible Trading Service. With faster speeds and more channels of electronic communication, regulators are feverishly trying to reign in trading transgressions and create more transparency. The most important acts are the US Dodd–Frank Wall Street Reform and Consumer Protection Act (D-F), which applies to any financial institution with operations in the USA; and EU MiFID (The Markets in Financial Instruments Directive 2004/39/EC). D-F is in ongoing implementation mode (despite some delays); the second generation MiFID II is a work-in-progress.
Most importantly, D-F requires swap trading companies to document thoroughly any deal within 72-hours if so requested by a regulator. That documentation must include all voice, mail and chat data relating to a specific deal.
Enter ‘big data’ and a 3-step implementation process.
The ability of big data products to handle large volumes of unstructured data is clearly the first step towards a highly scalable, near real-time monitoring of high-speed networks. The second step is developing applications that can rapidly synthesize the data and generate reports in the formats required by D-F.
Compliance applications are now becoming available to investment banks, institutional investors, hedge funds etc., from companies like Headstrong, Traiana and Fonetic. However, given that the fast-approaching drop-dead D-F compliance date is set to Q1 2014, the general state of market progress in the financial institutions, and in the regulators in major markets like the UK, France, and Germany, is too slow.
The third step is to develop applications that can spot non-compliant deals — for example, when a trader suggests on the phone or in an email that a deal is finalized ‘over lunch’, i.e. outside the range of deal-monitoring systems, and the monitoring system then alerts a compliance manager. This final step is still under development and is not yet a D-F requirement, but follows logically from investments in the first two steps.
Overall, there are improvements in transparency and critical analysis of trading activities on financial networks, but threats are still faced from other fast (but erroneous or malicious) information flows. On April 23rd, $130bn was temporarily wiped off the value of stocks in the S&P 500 after the Associated Press Twitter account was hacked and a false message stated that ‘Two explosions hit the White House. Obama injured’. The Dow Jones dropped 100 points in less than a minute. Today, the financial networks ‘move ’1000’s of millions of pounds in few seconds; but it is misinformation and lack of real-time compliance tools that can cost investors huge sums of money.