Delivering on-demand access to real-world market data, via a cost-effective service that can be implemented within a matter of hours, mdPlay enables quality assurance and stress testing teams to really put trading systems and algorithms through their paces.
When performance issues arise in today’s extremely fast and automated production environments, millions of dollars in trading losses can stack-up in a matter of minutes. In these situations, the pressure is on for engineering and testing teams to quickly implement a robust fix.
Responsible for locating and resolving the issue, then effectively testing and tuning debugged systems, these teams need to confidently assure the business that should the same set of circumstances reoccur their systems are guaranteed to respond with greater resiliency. However, if they can’t accurately recreate the situation, locating the root cause and determining if a fix has actually worked, without inadvertently impacting the performance of a downstream component, can prove extremely difficult.
Whilst many firms operate a duplicate of their production environment for quality assurance testing, many lack access to the actual data their systems were consuming when a particular incident occurred. This limits the testing team’s ability to effectively replicate the scenario, often because they:
These issues also limit how adequately new and enhanced trading software and algorithms can be tested before being released into production. Under MiFID II, investment firms will face demanding algorithmic testing requirements, including the need to ensure that trading systems and algorithms don’t ‘contribute to disorderly trading conditions, and can continue to work effectively in stressed market conditions,’ (RTS 6, Article 5, 4d). This is a direction being echoed by numerous other regulators globally and will necessitate firms to conduct periodic testing to ensure they can comply. The inability to effectively emulate realistic environments can lead to situations where the degree of testing that can be performed is disproportionate to the risk presented. Should unexpected market developments occur, unanticipated behaviour can then ensue resulting in market disruption, trading losses and regulatory fines.
Delivering the data necessary to accurately re-create real-world test environments for effective quality assurance and stress testing, mdPlay has been designed to enable firms to overcome these difficulties.
mdPlay provides testing teams with access to the actual market data streams received during any given trading period, so they can be quickly and easily replayed to precisely replicate the conditions experienced. The replays include all of the data that was being received simultaneously from every single venue the firm was consuming data from at the chosen point-in-time. All of these data streams are then replayed in a fully synchronised manner, with +/-20 nanosecond precision. Furthermore, to enable users to test how their systems cope under stressed conditions, customisable test scenarios can be curated where data is made available for re-transmission over pre-production networks at multiples of the originally recorded capture rates.
With this service, firms can continuously build-up a library of different trading conditions available for replay, using market data from any asset class. All of which are created on the client’s behalf by Velocimetrics’ dedicated professional services consultants. Firms can also take data they have previously captured (prior to deploying mdPlay) and Velocimetrics can then build this additional data into new test scenarios, so it’s also made available in the library for replay. Multiple users across the business can then digitally access these replays on-demand from a Playlist. They simply select the data they wish to retransmit and click play!
mdPlay is a highly cost-effective service that can be implemented within a matter of hours, with minimal resource and on-going maintenance requirements.
Retransmit real-world market data streams on-demand, for accurate testing
With mdPlay, users simply request the market data received during a specific time period, such as a certain segment of the trading day or the entire day, and it’s curated for them. The Velocimetrics team adds the file to the firm’s library and users can then immediately retransmit the data as many times as they wish with +/-20 nanosecond precision.
mdPlay offers the ability to capture and replay real market data streams, versus just synthetic or test data. In doing so, enabling subscribers to very effectively test how their algorithms and supporting systems will react when faced with the possible variances in the market data’s volume, content and rate experienced in production.
The retransmitted market data ticks are timed so precisely that even listening components are unable to distinguish between the original captured packed stream and that being represented. With a mdPlay capture, every single inter-packet gap is faithfully reproduced with nanosecond range precision, offering the data streams necessary for accurate testing.
Access completely customisable test scenarios
Clients subscribing to the mdPlay service can also request the curation of specific test scenarios for retransmission. The scenario might be, for instance, that the data received during certain periods on a specific trading day is manipulated so it’s replayed at various speeds, such as, 5 or 10 times that at which it was originally received.
Recreate precisely repeatable testing environments
With mdPlay users can stream files that enable market data events to be re-experienced as an exact replica each and every time the data is re-transmitted. This approach enables the same test simulation to be meticulously repeated time and time again, providing testing teams with the conditions necessary to prove without a doubt if an enhancement has, or has not, been successful.
Rapidly identify a problem’s root cause by accurately reconstructing the exact situation experienced
mdPlay enables subscribers to immediately request a copy of the exact market data feeds being consumed by their systems when an incident occurred. By retransmitting this data in their pre-production environment, firms can precisely replicate the situation originally experienced. This can be achieved within a matter of hours of the incident happening. In doing so, creating the conditions necessary for engineers to quickly identify why certain components reacted in a particular way and identify the problem’s root cause. This approach eliminates the need to trawl through log files or employ other time-consuming methods to achieve this goal.
Once a fix has been implemented testing teams can then consistently replay the market data necessary to recreate the original trading conditions. This provides testing teams with the confidence that should the same market events reoccur, their enhancements will ensure that the firm’s algos and trading systems respond in a more positive manner, limiting potential future trading losses.
Therefore, with mdPlay firms can reduce the time and cost associated with identifying a problem’s root cause and increase the accuracy to which solutions can be tested.
Address MiFID II’s algorithmic stress testing requirements
Under MiFID II, investment firms will be required to test the design and performance of their algorithmic trading systems, trading algorithms and strategies, prior to their initial deployment or substantial updates. The goal of which is to ensure that they don’t ‘behave in an unintended manner;’ (RTS 6, Article 5, 4a) or ‘contribute to disorderly trading conditions, and can continue to work effectively in stressed market conditions,’ (RTS 6, Article 5, 4d).
These tests are required to take place in non-live environments, therefore it’s going to be important for investment firms to be able to create and run realistic test scenarios in their pre-production environments. The type of tests that really put their algos and all supporting components through their paces. Enabling investment firms to record and replay with high precision the market data received during any given trading period, and request the creation of specific scenarios, mdPlay can help firms create the environmental conditions necessary for effective testing.
The MiFID II algorithmic testing requirements include the need for firms to annually self-assess their algorithmic trading systems ‘to ensure they are capable of withstanding increased order flows or market stresses.’ The MiFID II regulatory technical standards require investment firms to ‘design such tests to be specific to the nature of their trading activity and to their trading systems’ (RTS 6, Article 10) and that these tests include ‘running high messaging volume tests using at least twice the highest volume of messaging received and sent by the firm over the previous 6 month period.’ (RTS 6, Article 10, a)
To address this mandate, investment firms will need to identify and record ‘the highest volume of messaging received and sent by the firm over the previous 6 month period.’ and then put in place the capability to test that their systems can cope with double this capacity.
With VMX EndToEnd and the mdPlay service, investment firms can effectively detect, record and then re-transmit the highest volume of market data messages received or sent at twice the speed, or indeed any other rate the firm wishes. In doing so, enabling investment firms to very easily stress test their algorithms, and all supporting components, to meet this requirement and conduct effective capacity planning.
Furthermore, subscribers can request the exact replays they require to simulate market conditions, shaped by not only MiFID II’s regulatory requirements but also their specific business needs, instead of just being restricted to the testing scenarios provided by trading venues. Additionally, these market data feeds can be replayed on-demand, in a precisely repeatable manner, as many times as the subscriber wishes.
Eliminate the requirement for weekend testing on trading venues
Some trading venues distribute new feed formats for a limited time, during out-of-hours periods such as weekends. Thus enabling participants to test against these feeds. This often requires testing teams to work the unsociable hours these feeds are available.
With mdPlay, subscribers can capture and record the feeds and then replay them in their testing environment at a time suitable for them. This eliminates the need for weekend testing and removes time limitations, as teams can re-replay the data as often as they wish.
High-speed market data network analysis
With mdPlay, users can also understand if there were any underlying network issues impacting the data consumed during a specific trading period. mdPlay can accurately detect market data sequence gaps and generate detailed message rate statistics, achievable at sustained speeds of over 1.2 million packets per second, per feed.
This enables users to:
- Detect gaps in the market data they are receiving, indicating quality issues
- Gain a breakdown of the bandwidth of different feeds
- Calculate message counts
- Classify the message type
- And determine whether it’s the original packet or a requested retransmission that was sent.
Implementation within a matter of hours
By capturing and automatically discovering which market data streams are being consumed, the time, effort and costs associated with manually identifying which market data streams are present in an environment is dramatically reduced. This means that Velocimetrics is able to implement the mdPlay service so clients are up-and-running capturing data within a matter of hours. This data can then be quickly curated into replay files, available for retransmission into their test environment.
Furthermore, mdPlay continues to automatically detect the presence of market data feeds on an on-going basis, minimising the maintenance overhead.
The software required by firms to take advantage of the mdPlay service is deployable on servers as small as 2U. As such, mdPlay incurs a nominal datacentre footprint, limiting the financial implications of rack space and it’s associated costs.