Skip to main content

Part 2 of 3

Since the installation of first transatlantic undersea cable, trading technology providers were tasked with simple mandates:
  • Deliver information faster, from as many sources as possible and help investors and market operators to make better decisions and implement them quickly.
Fast forward to 2021 and the imperative for financial services firms of all ilk has moved away from just raw speed and toward the systematisation of trading decisions and the industrialisation of their implementation. Moreover, banks are still engaged in a struggle to regain competitiveness and bring down the cost / income ratios of their trading businesses. Therefore, the drive for automation and for the optimal allocation of human capital to the most value-added tasks now extends to all phases of the trading and client lifecycles.
This imperative – borne out of mounting regulatory constraints in the wake of the financial crisis – created an opportunity for technology vendors to alter their approach to systems architecture such that expenditure on both trading desk personnel and corresponding technology resources can be constrained or unleashed at will. Herein, the belief is that trading systems must be designed to preserve the ability of human traders to manually intervene in the operations of complex, automated execution programs when they deem it necessary to do so on either a pre-trade or at-trade basis, and that there is no other more optimal arrangement of functional capability – but, there is.
In this series of articles – produced in partnership with Itiviti Group – capital markets consultancy GreySpark Partners explores the current dynamics of the long-running debate over the dominance of high-touch versus low-touch trading systems design, arguing that the distinction between the two types is no longer necessarily relevant from an end-user perspective.

Buy & Build Modularity: The Best of Both Approaches

In 2021, the client-facing franchises of trading businesses need their trading technology and its underlying systems to provide three, non-functional capabilities – that is, efficiency, agility and velocity – to remain relevant into the future. This hypothesis means that investment banks and non-bank brokers must be prepared to service their end-investor customers’ needs with fit-for-the-future technology that leverages flexible production tools that are derived from a trader’s real-world practices and workflows. Specifically, GreySpark believes that trading businesses require technology capable of satisfying those three non-functional capabilities needs in order to:

  • remain relevant in a competitive and capital-constrained environment;
  • reconfigure the business at will; and to
  • unshackle change management and thus incentivise innovation thinking (see Part 01 of this series of articles).

Prioritising these trading technology needs means that businesses must allocate their capital intelligently by spending on systems research and development where it makes a difference to the franchise to do so, and to save on systems R&D where it does not make sense to spend. Hence, trading businesses rely on innovative approaches to systems design by third-party technology vendors to develop product offerings that intelligently bridge the gap in the R&D spending allocation and decision-making processes.

A current, leading approach to vendor-provided trading technology design that gained significant traction over the last three years is the so-called Buy & Build Modularity Approach founded on the idea of trading technology as platforms that contain functional components that be switched on or off at the user’s discretion. As such, Buy & Build modularity is optimal for trading businesses struggling with capital constraints because it combines the advantages of both packaged software and development frameworks by providing set functionality off-the-shelf while also accommodating a high degree of components customisation.

Vendor-provided systems design that incorporates the Buy & Build Modularity Approach is also optimal for capital-constrained trading businesses because it ensures a fully functional platform is available to users from the initial time of its development. Not only do the systems function from Day 1, but their architecture allows users to continuously expand or customise the system to create competitive differentiation for themselves in line with their business needs and available investment budgets. Because the system functions out-of-the-box, it thus permits competitive differentiation without creating dependency on the system’s core trading functionalities or on software development teams, whether they be in-house, vendor or certified third-party developers.

However, as the production software in use by the client diverges ever farther from the base software, a high quality of vendor-provided service and support is required. Where service and support lag, users can find that their customisations may be incompatible with newer versions of the system, resulting in the user having to ditch costly customisations or running outdated software that does not benefit from new developments on the part of the vendor.

Figure 1 depicts the evolutionary journey that investment bank trading business technology stacks, for example, typically undertake when the institutions work with their buyside clients and technology vendor partners to enable a Buy & Build Modularity Approach to front-office brokerage / exchange platforms, systems or solutions integration. At the core of the architecture and design philosophy is the convergence of high-touch and low-touch agency modules into a unified system that enables the best elements of both approaches with an optimised level of process and workflow automation throughout.

Herein, clients benefit from the ability to independently pilot a solution capable of supporting the specificities of how they manage flow or orders within their own firm – as well as how they prefer to execute – using a utility-like interface that masks the complexity of the systems underneath. In turn, the trading business benefits from the ability to reconfigure its instruments and products outlay at will, offering an overall slate of best-of-breed services that accentuate its niche capabilities and specialisations on a per client type or per marketplace basis.

A subsequent decrease in the levels of ‘friction’ between front-office trading systems used by clients versus any element of the trading business’ front-to-back technology stack is then realised, allowing personnel the resources required to either manage change effectively or to think innovatively about how best to meet client services technology needs over the medium-to-long-term predominantly by buying in – as opposed to building in-house – new solutions.

Figure 1: The Evolution of the Trading Business Front-office Trading Technology Stack

(click image to enlarge)

Source: GreySpark analysis

Asset Class Agnosticism: Trading Infrastructure that Works for All or Most Asset Classes

In 2021, GreySpark believes that trading business and technology vendor efforts over the last decade to build systems from scratch – or by using existing components – that are functionally capable of sleekly providing cross-asset or multi-asset class capabilities were not expended in vain. There are a handful of either in-house built or vendor-developed systems that, objectively speaking, work well for either cash equities / cash FX / vanilla structured products trading or for FICC trading purposes. However, no one system can provide the functional capabilities coverage required to allow traders to switch processes and workflows seamlessly between any or all asset classes – and their associated instruments or products, no matter how bespoke they may be – at will.

Arguably, cross-asset or multi-asset class trading systems design approaches from the outset were malformed in that they were:

  • Advantageous – Banks and brokers took advantage of opportunities to cohesively knit together the systems underlying trading platforms covering some, but not all, asset classes only when prevailing changes in market structure and the costs associated with markets connectivity organically incentivised them to do so; and in that they were
  • Demand-driven – Technology vendors, in struggling to respond to growing demand to aid their customer’s ambitions, began to focus less on building bespoke, trading desk-specific systems and began instead to focus more on interoperability middleware that creates containerised data management desktop environments.

As such, the lessons learned through the shortcomings of the new technology outputs produced since 2010 show that the markets connectivity costs rationalisation objectives of trading businesses are more readily achieved when they move away from the rationale that a cross-asset or multi-asset class trading system must achieve the possibly-impossible goal of covering all asset classes, products and instrument types all of the time.

Instead, trading businesses and their end-investor clients benefit from rationalised systems build or R&D capital expenditure more rapidly when the challenge is flipped on its head. Specifically, by focusing on the elements of the overall trade lifecycle value chain that are increasingly fundamentally the same for every trading business, franchises can more readily leverage vendor-provided products or solutions that address efficiency gains therein. Specifically, the elements of the trade lifecycle value chain where efficiency gains are most-readily achieved in 2021 include:

  1. Pre-trade Analytics – Where APIs and other, more dynamic market and trade data push / pull mechanisms and toolkits are now frequently used to visualise risk management information in a 360-degree, on-demand fashion;
  2. Client Connectivity – Where the increasing pervasiveness of API-centric, hub and spoke design thinking now means the end-investor client users of a brokerage solution can take full control of their liquidity flows and make markets competitively with the largest buyside firms and sellside institutions;
  3. Venue Connectivity – Where ultra-low latency market gateways and business intelligence software tools now combine to provide users with capabilities tailored to their businesses’ needed and workflows, regardless of whether they are native or FIX-based;
  4. Real-time Analytics Where specialised software toolkits that facilitate informed trading decisions now allow brokers the ability to infuse them with their own algos’ IP and market intelligence while simultaneously actively surveying high-touch and low-touch inputs to ensure regulatory compliance; and within the
  5. Post-trade Lifecycle – Where AI, machine learning and robotic process automation software is now frequently deployed across the settlements and reconciliations portions of the overall phase as a means of error elimination and to ease the burden of manual end-investor client reporting or regulatory reporting generation.

By focusing on those five areas of the trade lifecycle value chain that are agnostic as a result of how data is handled within them, trading businesses and technology vendors can simultaneously create efficiency and rationalise costs within the processes and workflows that must be performed in order to facilitate and service client trading activity. This approach is also more rational than any historical attempt to build an ‘all asset classes’ / universal trading platform that is gnostic to the specificity of the trading processes and workflows utilised by all types of traders globally. Simply put, the objective is to move away from solutions and systems design thinking that is predicated on the creation of cross-asset or multi-asset class trading functional capabilities because:

  • There are some parts of the trade lifecycle value chain that are the same whatever the asset or security being traded is, and then there are parts that are asset class-specific; but
  • Multi-asset class trading is, arguably, asset class-specific regardless of the inherent desire to build trading technology that services all possible investment and financial markets trading demands.

In reality, all trading activity outputs are predicated on the flow of financial messages – in all their multitude of forms – between systems; as such, trading activity outputs are not predicated on the ability of a system to successfully standardise trader processes and workflows in an optimal form for all traders. This means that a data management infrastructure-centric approach is optimal because – once costs savings within the agnostic elements of the trade lifecycle are achieved – trading businesses and technology vendors can then easily and more logically leverage innovative thinking from one asset class into innovations for trading in other instruments or products in either the same asset class or in other, separate asset classes. For example, many investment banks in recent years transferred the lessons learned from deploying cash equities execution algos into the building of likewise cash FX execution algos (see Figure 2). Those banks could efficiently do so because the data inputs and outputs underlying each markets’ infrastructure became inherently similar over time as the methods used to transact volume in any size shifted away from relationship-centric, high-touch, OTC, dealer-to-client environments and into algo or algo negotiation-centric, low-touch, e-manual, exchange or exchange-like environments.

Figure 2: Share of Investment Bank Trading Business E-trading by Tickets, Select Instruments

Source: GreySpark analysis

Automation of Trading Activities

Ultimately, an asset class-agnostic, Buy & Build modularity approach to trading technology systems design allows both trading businesses and technology vendors seeking to sell into them the ability the focus on the systemisation of human interactions with financial market data inputs and outputs. Specifically, this means building products or solutions from the perspective of:

  • An examination of the tasks that traders perform for themselves or for their clients, and the decisions that those individuals make, either consciously or unconsciously, on how to best execute on those tasks;
  • A divination and documentation of those task-specific decisions into a rules-based logic; and, finally
  • The implementation of that logic into software.

In this regard, the objective is not to further automate low-touch execution flow tasks; those tasks are already benefitting from levels of automation and their journey to full automation can be further improved, incrementally, over time. Instead, the objective is to focus on the areas of the trade lifecycle value chain and the tasks therein that remain subject to human decision-making, to question why a human decision is necessary in order to execute the task and to then document the reasoning.

This approach stands in contrast to the benefits brought by machine learning applications – which are dependent on a standardised set of inputs producing an expected set of outputs – because it involves getting the root of the reasoning behind what traders do to facilitate and service trading activity, and to the root of the reasoning why. Patterns in behaviour can then be extracted and systemised to the benefit of the trading business and its clients through the elimination of costly, time-consuming, high-touch activities that historically were the preserve of a trader’s so-called magic, thus creating digital locates for trader tasks such as:

  • short-selling;
  • portfolio pricing and hedging;
  • facilitation; and
  • execution advisory.

By bringing these historically high-touch activities into the low-touch fold using solutions and systems designed from the outset to accurately replicate human decision-making at key intervention points, trading businesses incentivise their clients to digitally interface to the fullest extent possible with the franchise’s technology overlay. This change would then create an opportunity for the businesses’ traders to analyse the sum of the picture of the data, or meta-data, created by all client interactions and to then perform new tasks designed to enhance the effectiveness of those client-generated outcomes.

In these ways, trading businesses can leverage data to generate information, and they can then use that information to make better decisions both for their clients and for the business. Arguably, the Industrial Revolution produced a world of sharp distinctions between companies and their customers. However:

  • because the pace of digitalisation continues to accelerate unabated, the nature of ‘work’ in industries such as financial services became modular or broken up into smaller packets that can then be farmed out to traders with ever-more specific skills sets.
  • This means that the trading business then becomes reimagined as a platform that synthesises modular packets of work to make ever-higher value products and services for the client base.

This democratisation effect is contingent though on trading technology solutions and systems designed to promote these new ways of working that depend on the de-siloisation of the wealth of information created by clients that can be used for real-time intelligence analytics purposes.

Investment banks and non-bank brokers now make wider use of either proprietary or open application programming interfaces (APIs) to allow both front-office sales-trading personnel as well as their clients to facilitate the construction of cross-asset or multi-asset cash equities and FIC business and trading models.

The result is a gradual convergence toward a single, agency trading business offering in which one P&L, one system and one team operate a low-touch model in which all the previously high-touch, repeatable tasks and workflows are automated away to improve the productivity of the humans managing both the desk’s operations as well as its clients needs. This shift reverses the historical state-of-play where the choices for low-touch trading teams and systems were subordinated to their high-touch brethren.

In doing so, new economies of scale can rapidly be realised that allow both broker-dealers and their buyside clients to zero-in on their niche agency market-making or proprietary trading capabilities, thus accentuating the value of the personal relationship elements of the broker / client services paradigm through the identification of low-touch client trading behaviours and patterns (see Figure 3). In this regard, the majority of investment banks are observed as commonly understanding that all of their institution’s data inputs or outputs – no matter how siloed they may be – must be viewed as an asset that requires continual curation and maintenance over time in order to avoid rapid levels of degradation and depreciation.

Figure 3: Client Trading Profiling Tools to Enable Real-time Benefits for Trading Business Lines & Technology Teams

Source: GreySpark analysis