Pivoting Fixed Income E-trading Technology Assets to Accelerate Compliance
In June 2015, a small group of quantitative traders working for the UK division of a Tier I global investment bank had a problem. The bank’s regulator was about to begin an audit of the institution’s algorithmic trading activities across several fixed income and currencies instrument types to check whether appropriate risk management controls were implemented to prevent a possible blow-up of market activity during sustained periods of shallow liquidity.
For the first time, the UK-based quantitative traders – who were providing pricing and making markets for US-based clients in European FIC instruments and products – would have to adequately explain how their black box trading strategies functioned. By 2019, fixed income algo trading among the Tier I investment banks active in European and US markets was commonplace, and the institutions were accustomed to reporting to their regulators on the specifics of what their strategies were doing and how the programs created to execute those strategies were achieving their aims.
Nonetheless, the parameters of two simple truths remained pertinent:
- Electronic Markets are Increasingly Susceptible to Temporary Trading Halts – For example, 8,896 circuit breaker trigger events were observed on 3,360 ETFs and stocks instruments traded in EU-based markets between April 2016 and December 2016 (the precise ramifications of these events on liquidity quality in EU-based FIC markets over an extended period remains unknown); and
- Many Bonds are Hardly Ever Traded – Giving rise to concerns from the EU’s European Securities & Markets Authority that the conditions that create so-called disorderly markets in non-FIC instruments could soon become commonplace within increasingly algo trading-centric bonds and swaps liquidity pools.
Enter ESMA’s December 2020 consultation paper review of algo trading activity within EU markets, the findings of which were set for finalisation and presentation to the European Commission by the end of July 2021. The onset of the review – which focuses heavily on the provision of direct electronic access by broker-dealers to their clients to algo-traded markets – forced many investment banks and other types of market-making and trading businesses to rush the development in 2021 of costly compliance projects designed to ensure adherence to MiFID II rules.
This GreySpark Partners article argues, however, that the costs associated with this work need not be expended in vain for investment bank fixed income trading franchises, specifically. Evolutions witnessed in recent years in the design and functional sophistication of technology vendor fixed income e-trading solutions – combined with an increase in the quality of the market data powering the algos plugged into those offerings – instead means that assurance of regulatory compliance from a connectivity perspective can now be an invaluable competitive differentiator.
‘If You Liked That Algo Trading Strategy, then You’ll Love This’
Whether it is the case that growth in algo trading in any one instrument type within an asset class begets liquidity fragmentation, or whether it is instead the case that liquidity fragmentation incentivises algo trading activity is a chicken-and-egg question for the ages. What is clear, however, is that – once established as a practice within a financial market – widespread algo trading activity requires two forms of fuel to burn brightly:
- An increase in the quality of available market and trade data; and
- An increase in the functional sophistication of the trading technology systems needed to manage the complexities of pre-, at- and post-trade processes and workflows for orders that can span dozens of markets simultaneously at any one point in time.
In global fixed income and credit trading, both conditions are now arguably satisfied from a Tier I to Tier IV investment bank broker-dealer algo trading perspective, according to a H2 2020 GreySpark survey of 139 different types of bank and non-bank bonds and swaps brokerage platforms and venues inclusive of a multitude of fixed income securities instrument types transacted across all four major geographies.
Among the survey’s many findings was the observation that, while the global fixed income market’s structure continues to disperse or fragment in line with historical changes first observed by GreySpark in 2013, notable changes at the margins are beginning to coalesce around recognition by the market’s many participants that tight bid / ask spreads and low-volume trades are the inescapable post-financial crisis ‘new normal.’ As a result, bonds and swaps markets liquidity fragmentation is predominantly characterised in 2021 by a growing number of brokerage trading platforms and venues that are altering or launching business models focused on all-to-all (A2A), multi-polar trading protocols that are – in large part – driven by increasing competitive structures for order and trade execution pricing.
Frequently, A2A fixed income brokerage business models are characterised by the ability of market participants to access liquidity in both corporate and government bonds within a single platform, as well as the instruments’ associated swaps (see Figure 1). GreySpark believes that the increasing popularity of this approach to liquidity consolidation by brokerage platform operators is a deliberate response to buyside firm and sellside institution user demand for bonds and swaps pools that capture and aggregate as much of their pre-trade and post-trade pricing, order and transaction execution data as possible, and at a granular level. The development of these capabilities as a unique selling proposition for brokerage venue operators is – in some cases – forcing incumbent platforms to re-evaluate the ways in which they historically facilitated buyside, sellside and end-investor client relationships.
In particular, the growing prominence of algo trading within the fixed income space leads GreySpark to believe that all brokerage venue operators will become specialists – to varying degrees – in the arena over the short- to medium-term. The resulting commoditisation of algo trading services will inevitably see the implementation of costs associated with garnering the capability to allow market participants to transact bonds and swaps at high frequency to come down, as was the narrative in the cash FX markets since 2009. In 2021, liquidity formation methods in the corporate credit market increasingly became unrecognisable compared to likewise methods commonly deployed in 2015.
This hypothesis is based on the observation that the proliferation of new trading technology resulted in more opportunities for buyside firms and sellside institutions to seek alternative avenues to filling the other side of an order / trade / transaction. The hypothesis is also based on the observation that access to the market / trade / transaction data underlying liquidity formation exercises became increasingly democratised by a variety of different brokerage entity types and other, specialised vendors since 2017.
For example, GreySpark believes that the provision by all manner of bank and non-bank brokerage venue operators to their bonds and swaps trading clients of access to deep and sophisticated market and trade data services packages is already a competitive differentiator in 2021. Not only are such ‘additional services’ now widely offered (see Figure 2), but the services are also deemed increasingly relevant by platform operators to the retention of user market share (see Figure 3), according the findings of the H2 2020 GreySpark survey.
As the prevalence and relevancy to bonds and swaps markets participants of brokerage platform and venue-provided market data services grew since 2017, so too did the prevalence and relevancy of other, ancillary additional services that were designed specifically with algo trading in mind. The translation, then, of these brokerage platform and venue business model changes over time into changes in market participant behaviour that demonstrate an established predilection in 2021 for algo trading is best witnessed in the number of matching methodologies offered by the different varietals of liquidity pool operators (see Figure 4) and in the extent to which users are turning to those order / transaction execution outlets for firm versus indicative liquidity or pricing (see Figure 5).
As bonds and swaps brokerage platform and venue operators expand the availability of their e-traded instrument and product offerings, they must become more adept at catering to the particulars of buyside and sellside sales-trader processes and workflows by continuing to innovate on new ways to facilitate liquidity matching opportunities. The need to remain innovative thus creates an incentive for the platform or venue operators to move away from simply creating efficient means of buying or selling fixed income securities, and instead it incentivises them to focus on the measurement of the speed at which order execution or transaction execution can be achieved without unduly affecting pricing as a means of maintaining the quality of underlying liquidity and its volatility.
Evidence that bonds and swaps algo trading by investment bank broker-dealers, specifically, is the focus of brokerage platform and venue operator innovation efforts comes with the fact that, despite firm liquidity becoming a central element of the characteristics of multilateral e-trading platforms or venues, the notion that buyside firms are reliant on CIB broker-dealer-provided liquidity from an indicative standpoint remains relevant. Specifically, platform or venue operators want market participants to have every opportunity to take advantage of firm pricing as it allows for trade execution electronically at the shown price – thus guaranteeing order execution – while also protecting the interests of liquidity providers in otherwise fragmented or loosely-regulated marketplaces.
Better Fixed Income Algo Trading, Through Technology
As the bonds and swaps order and execution management capabilities of the bank and non-bank brokerage platform and venue landscape evolved since 2013 to become increasingly algo trading-centric, so too did the functional capabilities sophistication of the in-house built and vendor-provided fixed income e-trading technology solutions commonly utilised by investment bank broker-dealers for at-trade order and execution management purposes. As such, GreySpark observes three crucial, overarching forces as the drivers of a new equipment era in Tier I, Tier II and Tier III investment bank fixed income business and trading models:
- A Flattening of Investment Bank Fixed Income Trading Revenues – GreySpark estimates that Tier I global investment bank bonds and swaps trading revenues fell from USD 51bn to USD 40bn between 2014 and 2019, as post-financial crisis regulation meant broker-dealers could no longer warehouse significant amounts of risk on their balance sheets for their buyside clients.
- The Implementation of Commission-based Agency Trading Business Models – The focus on agency trading resulted in front- and middle-office headcount reductions. This, together with a regulatory pressure to automate desk-based operations, incentivised markets-facing trade electronification expenditure.
- The Rise of Robust A2A Corporate & Government Bonds Liquidity Pools – In these new liquidity pools asset managers, hedge funds, institutional investors and some wealth managers are encouraged to utilise electronic matching methodologies to trade more directly with one another. This approach contributed to the further erosion of sellside revenue models that were dependent on the ability to manage time mismatches between clients’ buying and selling interests.
The impact of these trends as they emerged was the increasing divergence between Tier I, Tier II and Tier III investment bank utilisation of vendor-provided fixed income e-trading solutions, specifically. For example, GreySpark observed that:
- Tier I Banks – Between 2011 and 2020, these institutions undertook a slow and expensive replacement of vendor-provided solutions with in-house built trading, connectivity and pricing aggregation tools. The objective of this exercise was geared toward garnering an increased level of control over fixed income e-trading technology spend so that specific types of offerings could be developed to enhance competitive advantages in line with the specific characteristics of each bank’s respective buyside client base.
- Tier II Banks – Between 2011 and 2020, these institutions undertook the replacement of previously incumbent vendor-provided offerings with either new, like-for-like solutions or with alternative types of fixed income connectivity or order execution software designed to provide banks and their clients with an evolved variation on historical capabilities. The objective of this exercise was to take advantage of an increase in technology spend budgets within the bracket that were large enough to support a period of evaluation of alternative vendor systems but were not large enough to support a new generation of in-house builds.
- Tier III Banks – Between 2009 and 2012, these institutions adopted vendor-provided order management, execution management and connectivity solutions for the first time. Although technology spend budgets for those banks began to increase from 2012, in 2021, many broker-dealers remain reliant on the same incumbent vendors that were originally selected.
Thus, from an investment bank industry-wide fixed income algo trading perspective, the e-trading technology challenge arguably lies most pressingly within the arena of Tier II and Tier III institutions that have yet to settle en masse upon the functional capabilities blend – in-house built vs. vendor-provided – that best accentuates any given competitive niche in client market share.
Figure 6 shows the high-level framework of components contained within a generic fixed income e-trading technology stack commonly assessed by the investment banking industry in 2021 as ‘functionally necessary’ for the at-trade rigours of bonds and swaps order and execution management. At issue, then, is an understanding by Tier II and Tier III institution technology buyers or developers of how best to insert into this genericised stack the algo trading functional capabilities within any given component that, once implemented, can express the institution’s competitive differentiation and specialisation in a particular fixed income instrument type or regional market.
The first area of the genericised fixed income e-trading technology stack wherein GreySpark believes that functional capabilities sophistication for bonds and swaps algo trading could be enhanced is in the installation of specialised markets connectivity infrastructure. This specialised infrastructure – which would take the form of a direct market access (DMA) connectivity layer, a systematic internaliser (SI) / organised trading facility (OTF) liquidity aggregation platform supported by APIs, and a handful of market gateways – can be competitively differentiated for Tier II and Tier III investment bank business and trading model needs, specifically, by the presence of an external low code development platform that acts as a kind of functionality wrapper that envelops the sub-components (see Figure 7).
For Tier II and Tier III investment bank fixed income broker-dealers, GreySpark believes that the utility of an external low code development platform in support of the DMA connectivity infrastructure required to support algo trading activity is found most pertinently in the ability to reuse the coding required for market gateways development and maintenance. For example, in 2021 there are dozens of possible bonds and swaps execution outlets in the European marketplace housed in a handful of exchange platforms and non-bank brokerage venues.
Many of these execution outlets offer market participants the ability to connect through standardised FIX protocols, but there are also many other outlets that still maintain specialised, native fixed income connectivity protocols that present a challenge for overstretched investment bank trading IT teams to build to and consistently maintain. Specifically, investment bank trading IT teams must be willing to:
- maintain process disciplines around the sharing of coding classes or libraries among many different trading desks;
- manage a multitude of gateway connectivity code components, which makes code duplication inevitable over time; and
- be willing to trust that unknown code developed by another team is reliable, which they are not prone to do.
A low code development platform for bonds and swaps markets connectivity gateway development and maintenance would alleviate these investment bank trading IT team challenges by enhancing the modularity of the software that has already been developed. Instead of launching blocks of code in large packets, developers could implement each individual software component as an artefact that if registered in a centralised database or store, and then – with the aid of an automation engine – create new market gateways as needed / on demand by assembling artefacts based on the requirements of any given trading desk (see Figure 8).
Using Natural Language Processing and stereotypic behaviours inputs from developers, a low code development platform can reduce the complexity associated with an increasing number of market gateway code artefacts by using a coding automation engine, which manages a centralised artefacts database. Artefacts are then implemented in the form of libraries with a standard interface, which creates small black boxes that are easy to assemble, understand and use.
Crucially, from a non-functional perspective, once the code is written for any given market in any given connectivity protocol and the necessary gateway is constructed and implemented, then that gateway becomes a replicable item that can be rapidly duplicated when it becomes necessary for the investment bank’s fixed income e-trading desk to connect to any new market that uses similar connectivity protocols.
The second area of the genericised fixed income e-trading technology stack wherein GreySpark believes that functional capabilities sophistication for bonds and swaps algo trading could be enhanced is in the inclusion of a specialised algo trading services module and a regulatory compliance services module. Those modules – when supported by an analytics platform for price, risk, margining and collateral management, along with a graphical integrated development environment and a user control dashboard – would create a robust offering for Tier II and Tier III investment banks to lease to their agency trading buyside clients as a regulatory compliance-as-a-service solution (see Figure 9).
A scalable, high-performance, low-latency complex event processing engine would underlie the graphical integrated development environment or algo container that provides users with visualisations that create transparency around what a particular trading strategy is doing as well as windows to start / modify newly-added strategies that are automatically generated at run-time, thus alleviating the need to nominally develop and distribute additional code into the system. Meanwhile, the control dashboard provides users with an interface that can be used to:
- select trading strategies from an inventory;
- maintain execution / control of strategy instances;
- automatically generate strategy windows from strategy definitions; and to
- select execution sites / servers / cores for each strategy instance.
An E-trading Technology Stack for Disorderly Markets Testing Compliance
Returning, finally, to the regulatory compliance challenges posed in 2021 by the December 2020 ESMA review of the suitability of the MiFID II’s oversight of algo trading activity within EU markets, GreySpark observes three new, emerging trends within the sellside of the fixed income market that are likely to become particularly impactful on the competitive differentiation of Tier II and Tier III investment bank broker-dealer bonds and swaps e-trading franchises over the medium-to-long-term. Specifically:
- The Growth of Portfolio Trading – Wherein the increasing electronification of trading processes and workflows create upsides for Tier II and Tier III investment banks in their ability to deal baskets of fixed income instruments more efficiently, with hundreds of line items consisting of different durations, credit quality, liquidity dynamics and trade directions in a single, aggregate transaction.
- Swaps Bundling – Wherein the growing proliferation of desktop interoperability tools across both the buyside and the sellside of the fixed income marketplace means that bank and non-bank brokerage platform and venue operators are now incentivised on a regional level to enhance the functional sophistication of automated execution-to-clearing workflows that support multi-swap package transactions.
- Dealer-to-Client Repo Electronic Trading – Wherein most Tier II investment banks and some Tier III institutions are observed as already sufficiently maintaining the technology stack components required to electronify repo trading processes and workflows so that a greater number of transactions could be executed on multilateral venues globally.
The emergence of these three fixed income e-trading trends is arguably symptomatic of the success of investment bank broker-dealer adaptation to the fragmentation of bonds and swaps liquidity observed since the onset of the financial crisis and that fragmentation’s subsequent incentivisation of an increase in the quality of market and trade data services provisioned by bank and non-bank brokerage platform and venue operators that lend themselves to algo trading business models. As fixed income algo trading became more commonplace, however, execution capabilities across the participant landscape became more commoditised within the core instrument types such as G10 govvies or on-the-run investment grade corporate credit.
As such, Tier II and Tier III investment banks are challenged in 2021 to find new frontiers within the overall fixed income landscape wherein their ability to consistently supply pricing and make markets to different client type niches is already well established. Maintaining such niches over the long-term though means that the business and trading models of those banks must remain bulletproof from a regulatory compliance perspective lest they risk losing client market share to direct or larger competitors.
Bonds and swaps algo trading in the context of ESMA’s ongoing review of disorderly markets thus poses a conundrum for Tier II and Tier III investment banks. Simply put, those broker-dealers can either:
- Continue to play it safe in the commoditised main of the e-traded fixed income instruments and products universe where disorderly activity is commonplace, but easier to contain; or they can
- Risk new ventures in frontier e-trading environments where competitive differentiation characteristics can be readily expressed, but at the risk of holding direct responsibility for disorderly conditions if black box trading strategies suddenly go bad.
Here, the regulatory compliance services component of an investment bank fixed income e-trading technology stack geared for algo trading specialisations becomes key.
Such a component could provide users with an extensible, multi-asset framework for agnostically testing trading algorithms at arm’s length from the algo owners by emulating the protocols used by each trading venue such that the investment bank’s trading solution acts as if were connecting to a live fixed income liquidity pool (see Figure 10). Specifically, the component would do so by creating a fully abstracted framework in which different market participant behaviours and different disorderly market conditions can be created in a test environment, allowing users to configure market emulators, run test scenarios and produce reports for internal compliance or external regulatory purposes to demonstrate that the algos tested do not create or contribute to market disorder.
As such, the costs and efforts associated with regulatory compliance projects by Tier II and Tier III investment banks, bonds and swaps broker-dealers in either established or frontier, niche fixed income markets such as compliance with MiFID II’s Regulatory Technical Standard 6 – which established model risk management mandates for algo trading systems utilised in the EU – are not expended in vain. Rather, the technology-centric approach to compliance through reusable markets connectivity capabilities creates a new service line within any given institution that ensures algo-generated bonds and swaps price- and market-making by either the bank or by its clients is compliant out-of-the-box with all relevant regulatory mandates.