By Marcus Consolini, Managing Director, Head of Asia Pacific/Japan, Ullink
Nowadays, market participants face increasingly complex challenges when trying to assess their risk exposure across markets, asset classes and order flows. In the post-trade world there is a high level of integration of risk measurement and management processes, but in the pre-trade world there is certainly less. This is because many participants still have segregated silos for their different trading channels, such as DMA, DSA, Care and Dark Pools. Attempting to aggregate the pre-trade risks of each discrete channel, requires a strong understanding of their differences, the ability to install centralised and localised risk controls, and the expertise to manage technology and system integration.
Trading channels – the issues with vertical silos and risk exposure
Sell-side market participants that provide a broad range of services to clients continue to be faced with a major challenge. What kind of risks exist within distinct vertical trading silos and how should they be both individually and collectively managed? Traditionally, Chinese walls were erected between specific trading activities because risk, technology and operational demands differed. A lower compliance burden, often with prohibitive costs for running an integrated and centralised platform, and less sophisticated trading clients, meant aggregation across trading silos was never put in place.
However, in today’s markets, increasingly sophisticated clients are simultaneously trading across multiple silos, requiring market participants to manage risk in a single consolidated layer. The fact that the segregation of trading activities often results in increased risk exposure for the sell-side market participant and less flexibility for the trading clients, means that those who can implement an integrated and flexible aggregation layer should be able to retain and attract further business.
Specifically, clients are looking for the ability to trade across multiple channels with a market participant that can manage their risk limit across those various channels: essentially, providing one firm but fluid trading limit that incorporates all market and asset classes. This can only be achieved with a horizontal risk layer that sits on top of the individual vertical trading silos.
Centralisation and localisation – the issues with multiplicity and aggregation
While this horizontal risk layer would provide market participants with a consolidated picture of a client’s overall exposure across multiple trading channels, there is a further issue to address: what risks can and should be centralised and what risks must remain localised, for example whether at the exit point of an algorithm or a best execution engine, before entering a dark pool or a crossing engine?
In practice, there are really two types of risk assessment involved with many trading platforms offering multiple channels and strategies. First, there is what a specific client is actually doing, that is, what are the aggregated positions they hold across markets, asset classes and channels. Second, what is trading on the markets at any point in time through the sell-side market participant, that is, their full real-time market exposure. The first assessment gives a complete view of trading activity for all clients across all channels – effectively the client safety net. The second assessment looks at trades just before they exit the platform and hit the market – effectively the market participant’s safety net. The first assessment is where risk can and must be aggregated, while the second is where segregated systems dealing only with specific risks reside.
Business integration – the issues with different risk categories
Although aggregation across channels is required in order to accurately assess total risk exposure, the next challenge is how to set up the risk controls needed for each discrete channel. For instance, the risk parameters and validations when placing an order directly onto an exchange, for instance DMA, are and should be different to placing a basket of orders within a fully automated strategy. This often results in multiple trading systems provided by multiple vendors coexisting in the same environment.
When trading in a straight DMA channel, buy-side clients are the originators of orders and they retain complete control of the orders, with the ability to amend or cancel them whenever they want. Risk control in this scenario is relatively straight forward, as points of failure in this trading path are minimised.
However, the scenario becomes more complex within fully automated trading strategies, such as algorithmic trading. Buy-side clients neither retain full or direct control of their orders on the market, nor does the automated strategy placing those orders directly on the market have control. Rather the trader ends up with a handful of static parameters that are sometimes not even adjustable in real time. In the event that something in the algorithmic strategy goes wrong – and it has happened with alarming frequency during the past few years – it is not always guaranteed that the buy-side clients or the market participants themselves will have the ability to intervene and take control over the execution process. This highlights that the type of risk controls required compared with a straight-through DMA channel are different in nature and complexity and should remain segregated.
The problem is that buy-side and sell-side market participants have tended to adopt the fastest and, apparently, most efficient platforms and supporting systems to meet immediate competitive demands, without fully-understanding the risk assessment tasks that they might need to perform. The result is a technology stack that is a mix of in-house built and vendor-supplied systems. As both international and local regulation increases and compliance demands rise, obsolete platforms and architecture must either be completely re-built, re-engineered, or aggregated and consolidated. Hence, working with technology partners that are globally present and domestically locally focused is becoming increasingly necessary.
We’d love to hear your feedback on this article. Please click here
