QA Financial Forum Chicago | 9 April 2024 | BOOK TICKETS
Search
Close this search box.

Time-stamping trades and MiFID II — clocking up the options

flag-1569424186

 

MiFID II time-stamping will not be easy.

Big Ben. Credit: London Drum

 

When European Union legislators drafted the the second instalment of the Market in Financial Instruments Directive (MiFID II), they wanted trading firms to time-stamp trades to levels of accuracy in line with the speeds of high-frequency trading.

When it comes into effect in January 2018, the directive will require all trades to be time-stamped to Universal Coordinated Time (UTC) at a high-level of precision, and the directive will be enforced by the European Securities and Market Authority (ESMA).

Constant monitoring and testing will be necessary to make sure that the clocks used by trading systems do not start diverging from UTC.

How will firms do that? There is no vendor with a market-leading leading solution for time-stamping and synchronisation. And that is good news, according to Neil Horlock, Global Architect for Exchange Connectivity at Credit Suisse and co-chair of the FIX trading community clock synchronisation working group. “There are lots of vendors who are positioning to give solutions so we are seeing competition in the space,” he said. “There was a risk at one point that time-stamping would be tied to a monopoly or oligopoly position held by a small group of vendors, with everybody having to move to a specific way of doing things. I think the way things have turned out has been very positive for the industry as a whole.”

Horlock added that trading firms had also been concerned that ESMA might appoint or create a single an agency to test and validate time-stamping solutions. This hasn’t happened, leaving the field open to test consultants and service providers to compete for business.

Under MiFID II, different trades will require different levels of accuracy. HFT market participants have to be recorded at the 100 microsecond level precision (one microsecond is equal to one millionth of a second), while non-HFT algorithmic participants will have to meet the one millisecond standard (a millisecond is one thousandth of a second). Human trades, executed over the phone or online, will be recorded to a one second standard. Regulators say that firms can synchronise their clocks either using GPS or through feeds from national laboratories running high-precision atomic clocks.

Traders at banks and asset managers who are looking for time-stamping tools that meet MiFID II requirements have two options, according to Ian Salmon, market advisor at Accedian, a Canadian firm that delivers network performance services. First, they can timestamp through their servers. The difficulty with this is that firms running a large number of interacting applications will have to synchronise those apps and their different internal clock to make sure the time-stamping works. The alternative, said Salmon, is to time-stamp at the firm-wide network level to verify and then store the time of trade execution.

According to Credit Suisse’s Neil Horlock, financial firms will be able at least to choose a solution based on their business requirements and their size. An investment bank trading in different markets and in different asset classes could have one network device that combines the order-flow across all those markets and asset classes, avoiding the difficult task of having to coordinate different internal clocks sitting in different places on a  convoluted network infrastructure.  

Meanwhile, a HFT firm might go with a combination of precision time protocol (PTP) software running on their server to record when the algorithm decides to make an order, combined with a network interface card recording messages sent and received from a trading venue.

The choice is a good thing, said Horlock. And there is additional flexibility in the rules about the way time-stamping works. “ESMA doesn’t specify exactly at which point in the decision making process you have to time-stamp. So, for example, a firm might choose to capture the arrival of the FIX message as the receipt of the client order. But it doesn’t have to be that – it could be when it arrives to your FIX engine, or when it goes to your order management system. It’s the potential for different interpretations that allows everyone to find a cost-effective solution.”

But the solution is one thing. Making sure it works is another. MiFID II is clear that it is not enough to have access to an accurate time feed. The internal clocks have to be tested and monitored.

Victor Yodaiken, founder of the Texas-based FSMLabs, a company offering software that ensures firms’ internal clocks stay synchronised to UTC, recommended that traders set up trading servers with multiple clocks, and then run software that can look at multiple sources and notice any discrepancies. When any one clock starts to lose time it can quickly be flagged.

Making sure clocks stay synchronised requires accounting for a number of factors, explained Yodaiken. Even within the space of the building in which the clock is located, the physical distance the signal has to travel has to be factored into the calculation of accuracy,  as well as potential time lags created by the various devices between the receiver and the server that is running the trading software. “These are places where things can go wrong,” said Yodaiken. “Your package or your connection might start to slow down. A device could malfunction. These things happen at high speed and the regulations are stringent.”

As with all things to do with MiFID II, the major challenge is that some firms are far better prepared than others for the new rules. “Some firms had pretty good time stamping, but most firms either don’t have it properly set up, or assumed it worked better than it did,” said FSMLab’s Yodaiken. “ We have gone into places where people said they had very high level of accuracy and then you see something that hasn’t been working for 10 months and no one noticed.”