Remember me

Register  |   Lost password?

The Trading Mesh

Accurate Time Stamping

Tue, 14 May 2013 09:26:48 GMT           

The National Physics Laboratory (the NPL) held an interesting Q&A this morning on accurate time stamping. The presentation also involved the UK FCA and the (ex-)Chief Scientific Adviser to HMG, Sir John Beddington. The main topic of the discussion was what role Coordinated Universal Time (UTC) (of which NPL is the UK’s ‘keeper’) could play in time stamping for the financial markets.

 

The FCA’s concerns were that:

  • the markets and their regulators should have accurate time stamps in published data;
  • as many regulators, venues and market participants (including ISVs) were running their systems off the same clock.

 

Currently, the most accurate commercially available clocks are caesium or rubidium (‘atomic’) receivers/clocks. Sir John's particular concern in relation to the market infrastructure was that it could withstand, if required, the GPS system being interrupted, jammed or spoofed. GPS is vital to the long-term running of atomic clocks. The transmission of UTC down dark fibre is not so reliant.

 

NPL has devised a system (at the moment for UK-only infrastructure) where:

  • NPL provides its UTC down its own dark fibre cable to Telehouse;
  • at Telehouse, there is a back-up atomic clock and a box provided by NPL to certify the accuracy of the UTC (aBox);
  • venues, market participants access the Telehouse hub through dedicated fibre optic lines (or market participants access a venue's connected hub by fibre optic cable);
  • fibre optics owned and managed by the venues/participants individually;
  • cables need to be no bigger than 100Mb/1Gb;
  • each market participant and venue has a Box.

 

The delays in the cable network and its jitter are being reduced and NPL believes it is capable of measuring latency between two points (eg between Edinburgh and London) to ensure ‘accurate’ time stamping across the UK.

 

The idea is that each venue and market participant measures and stamps orders, cancels, amends and trades against the same clock (showing UTC) to within a margin of error. The uses of this would include (a) benefits for market participants in terms of market data measurement and strategy analysis, (b) orderly market and market resilience monitoring, (c) helping to build a consolidated audit trail and (d) monitoring for market abuse. The infrastructure is designed to be scalable, run off commercially-available kit, using PTP and other existing industry protocols. UTC is measured and kept by most major developed countries across the globe.

 

NPL and Sir John sought feedback from market participants and the FCA on the concept, design and any further considerations to be brought to the table.

 

In the Q&A, these considerations included:

1.       what 'margin of error' and number of decimal places to which to stamp trades etc would be acceptable and/or possible - somewhere between 1 and 10 microseconds was thought to be ballpark - it was noted that software and hardware involved was likely to introduce some inaccuracies and variable time delays (or jitter) which would make measuring or reporting time any more ‘accurately’ or on  more granular basis pointless;

2.       what was the likelihood of an international standard being applied and/or the technology being capable of being implemented across the EU and the US - cross-border discussions are in their infancy, but it does seem that the use of the technology (and regulations evolving therefrom) is limited unless there is a move towards an international consensual approach, particularly given the likely costs of implementation;

3.       is it really possible or practical to insist that all market participants have a Box and a dedicated fibre optic cable down which UTC information is pushed:

4.       NPL had designed the system with major connectors to the UTC information and Boxes being 'the banks' which did not take into account other direct or indirect market participants, so a broader requirement to have all market participants having a line and a Box had not been thought-through;

5.       there was some feeling that the Box and UTC information should be at a venue level only, although others thought that executing and clearing brokers should be included and it did not seem that anyone was particularly keen on having all market participants keep a line and a Box;

6.       a cost / benefit analysis had not yet been undertaken - whilst the benefits of the technology to a CAT might be more obvious (although query whether this is practicable under MiFID 2 and across asset classes), the benefits from a market abuse, market manipulation point of view were less obvious for a number of reasons, including:

a.       one of the important factors in monitoring for market abuse is the information available to a participant at the time of sending an order/cancel/amend, not the time at which the order/cancel/amend was received (or in the cases of an order or amend, filled) and whether this technology could solve for this problem given software latency within participants' systems and geographical-related latency is another question;

b.      who monitors and keeps the information - is it the regulators, the venues or the market participants (or all or some of them);

c.       we need to draw a distinction between time stamping's precision, resolution and accuracy, since they are not the same thing and on which will regulators and the industry rely;

7.       again from a cost / benefit analysis point of view:

a.       each line from a market participant to the venue or to Telehouse would need to be a dedicated 100Mb/1Gb line;

b.      given that atomic clocks rely on GPS, but GPS likely to go down irregularly, is NPL solving for a problem that is likely to make the time stamping of orders/amends/cancels/trades more resilient than the remainder of the market infrastructure and significant cost.

 

The viability of and interest in this project relies on a number of imponderables, particularly those set out above. Politicians and regulators also need to work out (a) with each other as global-as-possible requirements (at least from the UK’s point of view, a standard approach across the EU and hopefully the US) and (b) the uses to which the information will be put, before the issue goes much further. However, all attendees and the FCA noted that the likelihood is that once politicians and regulators reach a position (if not an agreement) on (a) and (b), then whatever requirements will be imposed are likely to be imposed in very short order (even perhaps as Level 2 implementation of MiFID 2). In other words, the time to make representations and/or enter into consultations (or to present industry best practice or alternatives) is now.

 

 

The opinions and writing contained in this article are of the author alone and do not necessarily represent those of HFTReview.com.

 

, , , , , , , , , , , , , , , , , , , , , , , , ,