Clock Synchronization: A Matter of Timing
As trading speeds compress into tinier and tinier fractions of seconds, it is becoming more difficult for firms and regulators to keep precise track of time, which makes it nearly impossible to accurately reconstruct the order of market events or to rely on trading algorithms that depend on historical time-synched data.
“There is an inordinate amount of human resource and effort that goes into matching trades when there are time sync issues,” Dr. Jock Percy, CEO of Perseus Telecom, told Markets Media. “As trading has become faster, the volume of missed trades or mismatched trades has increased.”
Perseus Telecom offers a High Precision Time service which enables customers to synchronize their time systems across multiple data centers, with accuracy as low as a picosecond, or one-trillionth of a second.
The service includes access to a National Institute of Standards and Technology (NIST)-certified GPS antenna as well as Network Time Protocol (NTP) and Precision Time Protocol (PTP) connectivity wto the Coordinated Universal Time UTC timescale.
“The challenge has been conformity,” Percy said. “We all know that the United States government has provided this service. What Perseus has done is build a technical and distribution relationship with NIST so everyone in the financial community can plug into it.”
The challenge of clock synchronization was raised at a recent Senate banking committee hearing on financial market structure, and the issue is one that Finra, the SEC, and the FCA are examining. While Finra OATs requires clocks to be synced within a full second or (1,000 milliseconds) it’s now necessary to time stamp orders down to the millisecond.
Time synchronization is essentially a data integrity issue. In order to make any sense out of the order in which transactions take place, the clocks on all those computers have to line up with each other.
For example, if two traders hit an offer on a contract on an exchange, and a question arises as to who actually hit the offer first, the counterparty on the other side of that trade may not have enough liquidity to cover both bids. “He’s saying, ‘Well, one of you guys is wrong,’” said Percy. “If one of the traders had High Precision Time, however, he could say, ‘I have a correct synchronized clock because I have a certification from my time provider.’”
Basically time synchronization means making sure that the clocks all operate off of the same time. “It can just an error of a few millionths of a second to completely destroy the ordering information. At that point, not only do the automated trading algorithms not work properly, but your internal controls don’t work properly and you can’t really validate what happened if something goes wrong,” ” said Victor Yodaiken, CEO of FSMLabs, provider of TimeKeeper, a high-precision time synchronization service.
One way to safeguard the integrity of time stamps is to track multiple reference time sources, so that the clock can be continuously cross checked against other clocks. The time stamp is then provided with a record of how well it matches secondary, tertiary, or deeper sources.
“Our most advanced customers are running sometimes with 3 to 5 high quality time sources, which they cross check constantly not just so that they can get the fail-over and the cross checking, but also that they have documentation,” Yodaiken said.
Featured image by Nomad Soul/Dollar Photo Club
However, the fund declines to provide it.
Algorithm leverages machine learning to find inter-listed liquidity.
Buy-side wants new and better ways to find block liquidity.
New algorithm generates IOIs to find chunks of stocks.
New algo uses conditional orders and IEX routing to help traders.