SEC Tightens Clock-Sync Mandate
In preparation for the Consolidated Audit Trail and to help detect spoofing and layering attempts, the U.S. Securities and Exchange Commission recently approved shrinking its one-second server-clock synchronization tolerance to a 50-millisecond window that will be measured against the time kept by the National Institute of Standards and Technology’s atomic clocks.
The SEC first announced its intention to tighten the window in late April when it discussed the next steps needed to advance the CAT rollout.
“We’ve been proponents of the 50-millisecond standard because it is sufficiently granular for regulatory agencies to improve their ability to root out disruptive trading,” said Jock Percy, founder and CEO of Perseus.
The 50-millisecond threshold also is a reasonable performance requirement for those market participants who outsource a portion of their existing trading infrastructure, he said.
The SEC’s move is seen by many as a step in the right direction, but several more steps are needed if the SEC truly wants to achieve its goals of accurately detecting spoofing and layering, i.e., bad behavior on the part of the shortest-term market participants
“Rightly or wrongly, this is about high-frequency trading,” said David Weiss, senior analyst at research firm Aite Group. “That largely is what people are discussing; that is what is in Michael Lewis’s book; and that is what IEX is about.”
A 50-millisecond synchronization window will do little in detecting high-frequency trades that occur at sub-millisecond, and even sub-100 microseconds, according to Weiss.
“There is a brouhaha over IEX’s 350-microsecond speed bump and the regulator is taking its clock-synchronization window down to 50 milliseconds,” he added. “I just don’t know.”
Instead of regulating a fixed performance threshold for clock synchronization, Weiss proposed scaling the performance windows in proportion to a firm’s fastest orders.
“If a firm is sending an order that takes 500 microseconds, maybe scale the window to 10 times that, or 5 milliseconds, ” he explained. “If a firm is not involved in HFT or algorithmic trading, such as a block trader, maybe there should be a floor of 50 milliseconds.”
Meanwhile across the Atlantic, the European Securities and Markets Authority has taken a much more granular approach in its clock-synchronization requirements.
In the latest draft of ESMA’s regulatory technical standard for MiFID II/MiFIR, the regulator plans to require firms that have gateway-to-gateway latency times between their trading systems and trading venues of a millisecond or less synchronize their server clocks to within 100 microseconds of UTC. For firms with whose gateway-to-gateway latency is longer than a millisecond, they will need to synchronize their server clocks to within a millisecond of UTC.
For those who fall under SEC jurisdiction, the new requirement goes into effect on August 15. Systems that already capture NMS and OTC equities transactions in milliseconds must be in compliance by February 20, 2017. Those systems that do not capture transactions in milliseconds have until February 19, 2018, to be compliant.
More on Clock Synchronization:
Trading turrets are set to go the way of the cassette tape.
Next shift may be from algos to machine learning.
Tailoring equity-trading strategies to client specifications is a work in progress.
The advanced analytics will provide real-time infrastructure monitoring.
ICMA calls for European consolidated tape.