Among MiFID II’s extensive new transparency requirements is a new set of technical standards that relate to time-stamping, message-latency, record-keeping and reporting that affects trading systems used by market participants of all kinds. While relatively straightforward on the surface, the regulation’s requirement for time synchronization across trading-related applications are presenting practitioners with some serious logistical and technological challenges.
The time-stamping of transactions for a range of functions, including best execution assessment, cross-venue monitoring, and algorithmic trading audit, is complex. Technically, putting in place systems to meet the synchronization challenge is an extension of what many firms already have in place to monitor their high-performance trading systems. But the scale of the new measures, requiring significantly more data points to record across activities in a broader set of asset classes, means firms will need to adopt strict governance policies to manage the infrastructure needed to calibrate clocks, synchronize time-stamps and collect the data required for compliance.
For an investment firm to monitor best execution, it must be able to compare timestamps between venues on a like-for-like basis. Take, for example, measuring order arrival against first execution on a venue, where the venue disseminates timestamps in microseconds. It would be impossible to rebuild the order book at time of arrival at the investment firm unless the firm also synchronized down to the same granularity as the venue. This requires use of clock synchronization, from a venue perspective, as well as from an investment firm perspective.
MiFID II sets out different levels of granularity of time measurement and stamping according to the activities of the entity under regulation. For trading venues, the requirement is specified in terms of gateway-to-gateway latency. This refers to the time measured from the moment an order message is received by an outer gateway of the trading venue’s systems, sent through the order submission protocol, processed by the matching engine, and then sent back, until an acknowledgment is sent from the original gateway.
For trading venues – exchanges/regulated markets, multilateral trading facilities (MTFs), organized trading facilities (OTFs), swap execution facilities (SEFs) and systematic internalisers – offering gateway-to-gateway latency of more than one millisecond, the requirement is for time-stamping to a granularity of up to one millisecond. For these venues, MiFID II allows maximum divergence from the benchmark Coordinated Universal Time (UTC) of one millisecond.
For faster venues with gateway-to-gateway latency of less than one millisecond, required time-stamping granularity is one microsecond, with maximum divergence of 100 microseconds.
For market participants like exchange members and their clients, the requirement again varies according to activity/function.
For voice trading and request for quote (RFQ) activity, firms are required to provide a time-stamping granularity of one second, with maximum divergence from UTC of one second.
For high-speed electronic trading firms, the time-stamp audit points must be captured on the route of an order message from its origination by a firm’s FIX engine, through order/execution management systems to the venue’s market access gateway. Here, the required time-stamp granularity is one millisecond, with one millisecond maximum UTC divergence for non-HFT firms.
For high frequency / algorithmic trading, the spec is for one microsecond time-stamp granularity and 100 microseconds’ maximum divergence from UTC.
MiFID II outlines a hierarchy of reporting obligations. From a time-stamp perspective, these range from voice recording and quotes for OTC instruments governed by Article 16 and are subject to latency divergences of up to one second, through to Articles 6, 7, 10, 11 and 26, which govern trade reporting (near real-time reports to an exchange or APA, and transaction reporting (more detailed reports, often with T+1 delivery), which in extreme cases are subject to microsecond accuracy requirements. Finally, Articles 17 and 25 cover record-keeping by investment firms, trading venue operators and systematic internalisers, with the requirement to store a full audit trail for five years (and in some cases up to seven years).
To meet this array of requirements, firms’ connectivity solutions need to synchronize their business clocks in order to get the correct timestamp on each trade and on all order book data. Firms will require a comprehensive latency monitoring capability to detect any anomalies in the performance of the market data flow. It’s essential for firms to eliminate any jitter, or dissemination latency, from their information delivery systems, to ensure that the real-time view of the entire market at any given point in time is not distorted.
As such, MiFID II’s time-stamping obligations necessarily involve calibrating the clocks that measure activity on all internal systems involved in the trading process. This requires a compliant trading architecture that provides accuracy throughout the full order lifecycle: from trading application to market access gateway to exchange gateway to matching engine and back. In this way, firms need to generate a complete record of every new, amended, cancelled, partially traded and completely traded (filled) order.
Trading firms and venues will need to design their time synchronization capabilities at the one millisecond level as a bare minimum. What’s clear is that a piecemeal approach to addressing the time synchronization issue won’t suffice. Firms need to take a holistic approach if they are to avoid the specter of non-compliance, and the financial and reputational penalties that involves. Moreover, a holistic approach can bring substantial operational benefits.
Accurate time-stamping can underpin firms’ responses to the EU Market Abuse Regulation (MAR), to best execution provisions of other regulations like the US’s Dodd Frank, and other compliance-related issues. It can also yield business performance metrics unavailable till now, offering insight into operational efficiency, profitability and risk management.