Tracing Time: Clarifying MiFID II/MiFIR Timestamp Requirements
By Victor Yodaiken, FSMLabs
Originally published on TABB Forum
Recent ESMA guidance clarifies time-stamping and clock synchronization requirements under MiFID II/MiFIR, making it clear that regulators are not interested in running an experiment and are not willing to settle for window dressing.
At the end of 2015, European regulators issued wide ranging “guidelines” on the new MiFIR rules that show that, when it comes to time-stamping and clock synchronization, regulators are:
- Not interested in running a science experiment. They want useful data on trading records based on existing technology. In particular, they reiterated (against quite a loud marketing effort to the contrary) that GPS and GNSS time sources are acceptable alternatives to feeds from the national timing labs.
- Serious about timestamp data integrity and not willing to settle for window dressing. There was a rather sharp reminder that the existence of a high-quality time source somewhere in a trading facility was nowhere near enough – time has to be synchronized at all points, and market participants need to document and monitor their end-to-end time synchronization technology.
The draft standards had already made it reasonably clear that satellite time sources were acceptable – and the cost-benefit analysis was particularly emphatic on that point. But there was an energetic campaign by a number of parties to convince the markets that not only were direct or somehow “authenticated” feeds from the national labs the only acceptable sources of time, but that connecting to one of those feeds was sufficient to meet time synchronization requirements.
The recent guideline addresses both of these points head on. First, there is an unambiguous declaration:
“The use of the time source of the U.S. Global Positioning System (GPS) or any other global navigation satellite system such as the Russian GLONASS or European Galileo satellite system when it becomes operational is also acceptable to record reportable events.”
Following this is a warning that should apply to other time sources as well: that just having an accurate time source is not sufficient. Accuracy is needed at “any point within the domain system boundary where time is measured”:
“For the purposes of Article 4 of the MiFIR RTS 25, for users of a satellite system, even though the first point at which the system design, functioning and specifications should be considered is on the receiver (e.g. the model of the GPS receiver and the designed accuracy of the GPS receiver) used to obtain the timestamp message from the satellite, the accuracy required under in the RTS shall apply to any point within the domain system boundary where time is measured.”
It is made abundantly clear in the rest of this guideline (and in the draft regulations) that regulators want a comprehensive tracking of when decisions were made and when data was received or sent. Such tracking depends on end-to-end data integrity, and time is now part of the data that has to be right. Just having a high-quality satellite or terrestrial time source feeding into a data center or a rack of computers is nowhere near sufficient to assure that the computer servers that receive customer orders or issue orders to trading venues, or the counterparty server computers inside the trading venues are using reliable timestamps.
Even if one had a super-accurate time-feed straight from a cesium clock in a national lab coming into a data center, there is still a long path between that feed and the software that is generating trading operations. And sometimes feeds fail. Sometimes the switches and routers in the network can fail or introduce timing errors. Network cards, cables, operating system software, and time-sync software can all introduce errors.
Market participants that want accurate time at “any point within the domain system boundary where time is measured” need technology and procedures to test, to alert, to fail-over if possible, and to otherwise monitor time synchronization. The guidance shows that ESMA regulators have been thinking about that path and will not be satisfied with “we pay for a time feed” as evidence of meeting the standard:
“Operators of trading venues and their members or participants shall establish a system of traceability of their business clocks to UTC. This includes ensuring that their systems operate within the granularity and a maximum tolerated divergence from UTC as per RTS 25. Operators of trading venues and their members or participants shall be able to evidence that their systems meet the requirements. They shall be able to do so by documenting the system design, it’s functioning and specifications. Furthermore operators of trading venues and their members or participants shall evidence that the crucial system components used meet the accuracy standard levels on granularity and maximum divergence of UTC as guaranteed and specified by the manufacturer of such system components (component specifications shall meet the required accuracy levels) and that these system components are installed in compliance with the manufacturer’s installation guidelines.” (emphasis added)
The takeaway lessons are, at least so far, that complying with the new regulations is going to need some work, but nothing miraculous. And buying a GPS clock or a terrestrial time feed is something that should happen in the context of a comprehensive plan.