Corvil, which provides data analytics for electronic trading, is launching a service to allow firms to independently verify that they are meeting the new time-stamping requirements required by both European and US regulators.
The firm has announced the launch of a new solution that establishes a system of traceability to Coordinated Universal Time (UTC), which is critical for monitoring, analyzing and reliably reporting activity on electronic trading networks. The UTC Traceability Solution automatically sends alerts when trades to not meet the clock synchronization requirements, and evidence of this monitoring can be provided to regulators.
David Murray, chief marketing and business development officer at Corvil, told Markets Media: “The service uses multiple sources to independently monitor timestamping discrepancies and sends warnings in real time. Firms have done a lot of work on deploying infrastructure for clock synchronisation but it may not always perform as they expect.”
MiFID II, the regulations covering European Union financial markets, will be implemented on 3 January 2018. Trades need to be timestamped to UTC with a high level of precision, for example, within 100 microseconds for high-frequency trading. Non-compliance risks fines of up to €5m ($5.8m), or 10% of global turnover.
In addition, both the European Securities and Markets Authority and the US Securities Exchange Commission will soon require trading venues and market participants to have a system of traceability to UTC in place for all reportable events within the trading network.
Research last month from the National Physical Laboratory, the home of the UK's national timescale, found that many UK financial services firms may fall short of compliance with the MiFID II timestamping requirements.
NPL’s survey, covering 200 responses, found that more than half (56%) of firms used Network Time Protocol-based Internet Time for timestamping which is only accurate to the tenth of a second. NPL said in a statement: “While able to accommodate the requirements for human trading, such as over the phone and online, it cannot remain a solution for HFT and non-HFT, as these require 100 microsecond and one millisecond accuracy respectively.”
There is also continued reliance on GPS for timestamping by 14% of financial services professionals, despite the majority (79%) of those experiencing issues doing so, such as drop out, loss of accuracy, lack of synchronisation and leap second issues. “GPS can be vulnerable to jamming and loss of signal, and can only ensure compliance with continual monitoring and calibration to ensure that it is traceable back to UTC,” added NPL.
The report added that the ideal method to deliver a precise time signal, guaranteed to meet the accuracy required for MiFID II, is UTC delivery by fibre from a national timing institute, which was only being used by a fifth of respondents. NPL operates two atomic clocks which are accurate to one second every 158 million years, and is developing the next generation of optical atomic clocks which will be accurate to one second in 14 billion years. NPLTime provides a precise time signal directly traceable to UTC, which is independent of GPS, and complies with MiFID II.
Murray continued: “As artificial intelligence explodes in electronic trading, accurate timestamping is critical to be able to understand what has happened and reconstruct events in trade life cycles.”
In a recent blog Murray said time has always been a key factor understanding of causality.
“Something that happens after something else can’t possibly be the cause of it – right?,” he added. “ “This is highly applicable in the daily discussions of cyber security where machine learning is increasingly employed, and in many other industries as oversight discussions on AI amplify.”
Humans can react to events in one or two seconds but machines can execute entire transactions, consisting of millions of instructions and calculations across multiple software programs, in 20 microseconds or less.
“The machines surely find our glacial pace laughable,” said Murray. “The equivalent volume of data of the entire Library of Congress is traversing corporate enterprise networks in seconds to minutes, every hour of every day. Think about it: in the time it takes to say ‘synchronize your watches’ computers have executed millions of transactions."