MiFID II definitions for time-stamping and rules for record keeping to support tracking of trade orders — as defined in Article 25 of the directive’s regulatory technical standards (RTS 25) — even less than a year away from the deadline for compliance with the regulation, appear to be little understood, according to experts who spoke in a webinar hosted by Intelligent Trading Technology on March 16.
A first step in time-stamping is having an “accurate, sequential understanding about time,” said Ian Salmon, director at Ignite G2M. “The regulation introduces the concept of traceability and the fact that it has to be accurate to a particular clock rather than just accurate within itself,” he said. “Your latency across a network matters because it’s deeply rooted in what the MiFID II requirements are trying to address — putting together a consolidated view that can be referenced for a particular time.”
Certain trading events need to be better defined, including initial client order receipt and initial decision to deal, according to Bob Mudhar, partner at Citihub Consulting. “Where do you receive the initial client order? What does the initial decision to deal mean? When is a deal made? There are a lot of vagaries,” he said. “We often see firms overdo the requirements and think everything in every part of the stack from the first time the order hits outer edge routers to when it’s released to a venue has to be set to the same standard of time stamping and clock sync. … If you read MiFID II carefully, I don’t think it does.”
Standards and triggers
The technical standards for time-stamping that firms will have to meet under MiFID II include choosing the clock that will be used as a reference, and indicating the type of organisations involved in a trade, the type of trade and the level of granularity of the time stamp (i.e. microseconds or nanoseconds) required, observes Salmon. In that context, if a trader is contending with a dual listed stock, such as Apple on both Nasdaq and the Frankfurt Stock Exchange — an example cited by a webinar listener — “the requirement for cross-border is everything that’s done within your own infrastructure has to be documented well and time-stamped at the point of order submission,” said Salmon.
High-frequency traders, in particular, are concerned about three kinds of triggers, according to Mudhar. First is latency reducing infrastructure, such as proximity hosting, co-location or high-speed direct electronic access. Second is automation and third is message rates, which are vague, added Mudhar.
“What seems like a reasonably good definition on paper can be picked apart. Firms need to decide if they want to be high-frequency traders or not,” he said. “If you can’t measure your message rates and decide whether you want to have latency reducing infrastructure, you might decide you’re opting in. What that means in practice is it’s clearer if you’re a high-frequency trader what you need to record and the precision needed around time-stamping.”
Overall, RTS 25 can be approached a few different ways, but some have their drawbacks, according to Dermot Harris, senior vice president, regulatory solutions, OneMarketData, which sponsored the webinar. Firms may spend a lot of money and effort retooling servers to make clocks more accurate, even if regulators will not have as much use for that detailed information. “It’s much less likely that a regulator will do a deep dive into all those details unless a specific regulatory question or examination comes up,” Harris said.
Another method that firms use is packet-sniffing networks, capturing the packets that have accurate time stamps and collecting them in a data lake or repository, but, as Harris said, “it just punts the problem forward to the next step, which is making sense of all this raw data, when the only real thing you have any more is a time stamp.”
The third and “smartest” approach is “capturing precise and accurate time stamps at a few specific checkpoints,” said Harris, “and figuring out what the key decision points are, like making a routing decision, an outbound gateway and an order arrival gateway — capturing accurate time stamps at those points.”
While changing a gateway to improve accuracy by a “quarter microsecond” can be “a major disruption or cost change,” said Jim Northey, co-chair of the FIX Global Technical Committee, “we’re all underestimating the amount of work to be fully compliant with RTS 25.” The several hops that orders make in systems including messaging platforms, smart order routers, execution management systems and order management systems all have to be synchronised with same level of precision in time stamps, added Northey.
The understanding of and attention to RTS 25, as Northey and colleagues described, also was evident in the results of an informal poll conducted during the webinar. When audience members were asked if they are prepared for RTS 25 requirements, only 16% said they were, and 47% said maybe.