Liquidnet: What We Learnt At TradeTech
Industry participants met in Paris to look beyond MiFID II and Brexit. The changing buy and sell-side relationships, retaining access to liquidity, systematic internalisers (SIs), and periodic auctions were annual topics still up for discussion; however, there was a shift in focus this year towards data, machine learning (ML) and artificial intelligence (AI) as the buy side increasingly focuses on improving execution outcomes in a shifting market eco-structure.
Sourcing liquidity: looking for innovation
With the buy side providing the sell side with only partial information on the flow it is looking to execute, the responsibility to source liquidity in an evolving market eco-structure increasingly sits with the buy side alongside the obligation to provide best execution to end investors. Understanding where liquidity is pooling and retaining access to that liquidity remains a major concern for the buy-side. Unbundling has irrevocably changed the traditional buy and sell-side relationship, with nearly 60% of the audience highlighting the impact of systematic internalisers and electronic liquidity providers (ELPs) as the new providers of liquidity.
Nearly 40% noted the impact of changes to the market eco-structure post unbundling. As recently highlighted in the Liquidnet report on research unbundling¹, the separation of research from execution services has moved the ownership of information from the sell side to the buy side. For bulge bracket banks to continue to provide execution services for clients, firms need to demonstrate their ability to innovate to differentiate their execution as well as their research offerings. As more firms move towards systematic routing based on quantitative performance metrics, how brokers are selected, monitored and retained by the buy side is changing. Although 2018 saw the growth of new types of trading venues such as systematic internalisers and periodic auctions, the overall proportion of liquidity migrating to those venues still remains less than the activity in the closing auctions and has yet to significantly impact the percentage traded in lit continuous markets². In the first quarter of 2019, periodic auctions represented only 2% of total market volume and SIs’ market share had stabilised around 15%³. Given the current ESMA review on frequent batched auctions as well as forthcoming amendments to SI regulation, these market share percentages are likely to change as liquidity formation adjusts accordingly.
While liquidity is not yet shifting back to the continuous lit market, it is being repackaged as the sell side connects to new sources of liquidity in a bid to deliver enhanced execution. Given the rise in use of systematic routing and algo wheels, investment in technology is essential as the industry focuses on both the implicit and explicit costs of execution. The difference today is how firms are shifting from historic proprietary models of execution to a more collaborative open architecture, facilitating partnerships between different liquidity providers and vendors that would have seemed unlikely a few years previously. Given the current saturation of execution brokers in the market, it appears there are considerable hurdles for new entrants to gain access, but new partnership opportunities exist for those able to offer differentiated liquidity. This is not only for those bulge bracket banks investing in technology, but also regional and specialist brokers with local insight and access.
With the regulatory emphasis on improving transparency and accountability, future disruption and innovation in European capital markets is likely to come from increasing use of data and analytics, now considered necessary to measure execution outcomes. An audience poll showed that one of the biggest changes in trading strategies since MiFID II has been the growing focus on TCA as the buy-side continually seek to understand brokers execution strategies and focus on implicit as well as explicit costs of trading. The requirement now is for solutions which provide the buy side with a truly independent means of analysis alongside analysis provided from execution partners.
Focus on SIs
After a steep decline in the second half of 2018 and early 2019 where the amount executed on SIs decreased to an average low of €6.1bn per day, in March SI activity picked up again to reach €10.6bn. The previous drop in activity coincided with a continued fall in the average fill size to ~€18.0k in January, rising again to €30.5k as activity increased at the end of the quarter⁴. However, while overall volumes are stabilising, the mix of traded volume between ELP SIs and bank SIs appears to be shifting as some firms are becoming more comfortable with ELP liquidity. While only 42% of the audience agreed that ELPs were adding true liquidity differentiation in the market (see exhibit 3), some panelists suggested that ELP activity now represents as much as 50% of overall sub large-in-scale (LIS) addressable liquidity.
These new statistics contrast with last year when 41% of buy-side firms interviewed by Liquidnet highlighted using bank SIs only and 6% preferred not to connect to SIs at all⁵. As access to traditional risk capital becomes more constricted, buy-side firms are having to extend their execution relationships beyond bulge brackets – with traditional sources of liquidity including central risk books, agency and algos – to trading with ELPs directly. ELP SIs are also proactively changing their interaction by offering bespoke feeds to individual buy-side clients, as well as acting as liquidity providers within bulge bracket execution capabilities.
Industry participants still remain unsure as to how the mechanics of different SI models work, with 82% believing they have insufficient information on individual models. The expectation was that as more data becomes available from RTS 27 reports, market participants will be able to assess execution quality and fill rates achieved when routing to an SI, creating more confidence through increased transparency, particularly in relation to a better understanding of the mechanics of the ELPs construct. Proponents of ELPs argue that they offer an unconflicted pure risk model and it is the bank SI where more clarity is required between addressable and non-addressable liquidity and quasi agency flow. More industry work is underway to clarify the different liquidity flows within bank SIs and enhance RTS 27 reports but this will take time to filter through. As a result, 50% still see their biggest challenge as understanding who they were interacting within an SI and 57% of participants are unsure of how ELP SIs differ from each other.
Lack of clarity around SI models may well continue in the short-term given the forthcoming European Commission’s amendment to the regulation to subject SIs to the tick size regime for standard trade sizes. Panellists were split on the impact of this; some claiming this change will restrict the SI’s ability to price improve sub-LIS, potentially negatively impacting execution costs for end investors as ELPs would continue to focus on speed of access and tighter pricing for participants to ensure likelihood of execution. Others suggested that any change in subStandard Market Size (SMS) trading could lead to ELP SIs increasing their average execution size to meet large-in-scale thresholds, further altering the liquidity formation in non-bank SIs.
Data, data and more data: what does the industry need?
Increased competition in execution can offer improvements as new trading protocols and venues emerge. However managing different execution outcomes in a more fragmented trading environment requires continued investment from the industry in both technology and the provision of accurate data. The demand for more standardized, readily actionable datasets is underway as the buy side seeks to independently optimise decisions in portfolio construction, venue selection, execution strategies and timing.
As liquidity fragments over a wider range of execution options, buy-side firms must now collect data from multiple sources to analyse and enhance their venue and broker selections. Asset managers need to quantify how and why they are connecting to venues, requiring better preand post-trade analysis to predict execution outcomes according to the venue and routing mechanism selected. This is changing the dynamic between buy and sell side; those execution providers that maintain a close relationship with their clients to understand how they unwind positions as well as where their data and tech needs lie can develop an evolving reciprocal partnership between the buy and sell-side beyond traditional brokerage execution. While this could potentially lead to a concentration of business amongst a smaller number of bulge-bracket liquidity providers, participants on the market structure panel “The Paradigm Shift” noted the need for the sell-side to focus on execution quality will lead to firms branching out to alternative means of sourcing liquidity as they focus on providing a differentiated service . Given the cost pressures the industry is now under, firms are continually rethinking how they access and provide liquidity as they look to maintain margins through maximising operational efficiencies.
The Resurgence of the Consolidated Tape
This shift from historic proprietary models to more open collaboration between alternative industry participants is also leading to a resurgence in demand for a consolidated tape. The quality of data collated by ESMA has been gradually improving post MiFID II, enabling the industry to consider the next steps in the development of a European consolidated tape. Yet, some still caution against its implementation given the risk of wrong conclusions being drawn from inaccurate data. SEC Director Brett Redfearn acknowledged the current limitations of the US Tape and its planned review, including concerns regarding the costs of accessing data. The consolidated tape also requires time to aggregate data from different sources into one data centre, creating a disadvantage for those unable to pay for direct feeds.
Estimates regarding the industry cost to purchase data are in the region of $35.5bn at a 40% compound annual growth, with data becoming a disproportionate part of exchange revenues. The European regulator reiterated the regulatory requirement for a basic level of free data as soon as technologically possible across all asset classes. Any increased cost in market data can limit firms’ ability to benefit from data in their investment decisions. With both sets of regulators as well as the European Commission outlining plans to improve access to data, the challenge for the industry is to define what data should be free and readily available to all vs. that which should be paid for.
The Rise in Alt Data
While access to data aids execution and venue analysis, data is also a means of alpha generation in portfolio creation; in particular the growing use of alternative data and natural language processing (NLP) in interpreting the data. The rise in unstructured datasets, such as transcripts from social media or earnings calls enables asset managers to maintain an informational edge that was previously unavailable. One panellist on “how can leveraging alternative data sources help you make more profitable business decisions and who are the companies offering true innovation?” noted that in the US the demand for non-US data is disproportionately high in alt-data vs traditional datasets such as foreign language news sites. But alt data can also include internal datasets which are not accessed currently despite holding valuable information in improving workflow processes. The ability to gain insight from everyday transactions is the first step to independence for many buy-side firms in establishing new investment opportunities, often more relative to their organisations, as opposed to a sell-side analyst passing on an idea. The potential alpha of an individual workflow signal will depend on the firm in question, the timing and the potential market response to the information as opposed to a generic investment idea.
Optimal investment opportunities will continue to rise from a growing partnership between alternative and traditional datasets creating unique value-add for asset managers. Unbundling research from execution will transform the industry’s understanding and use of research by including alternative datasets in the definition of research as well as execution. Successful investment strategies will require buy-side firms to determine how and where to leverage data, in-house or through external vendors, and then how to best purpose these datasets within workflows to gain actionable insight. It is not only accessing data but the ability to extract value.
Another area discussed was the growing challenge in how to continue to access emerging datasets. Regulations such as GDPR and new regulations around ESG complicate how firms can access data from another legal jurisdiction and the proliferation of data sources will only make the process more complicated. Managing regional or local legislations within a global investment framework will become increasingly important – a hedge fund in Asia could trade in both Europe and the US, incorporating anything from local California legislation to both US and EU GDPR. While intelligent tagging may be a solution, only regulatory cooperation between jurisdictions will provide the legal framework for firms to access cross-border data successfully.
The role of automation and AI in the investment process: how to use it?
In recent years, growing amounts of resources have been dedicated to advancing automation of execution, including ML and AI. While the potential use of AI in financial markets and its benefits continue to be raised (see Exhibit 7), its use remains limited. Automated execution of portfolios or the transcription of voice trades into an electronic trail are two examples where AI could improve execution. However, to advance its usage, firms will not only need to commit financial resources but also address necessary cultural change within organisations. This will require AI specialists to work together with fund management and execution teams to fully understand current processes and develop optimal solutions that will help run the business more efficiently. Yet, a majority in the audience involved in trading and execution services acknowledged they are still not engaged in the use of AI tools, which could present a risk in firms developing tools that do not fit the traders’ needs.
The rise in AI and machine learning will not automatically replace the role of the trader. At its current stage of implementation, use of AI and ML appears limited to an efficiency play to ensure traders can retain access to the level of information ahead of placing a trade. Rather than relying on a human to “know” the market, the increase in available data together with the need for quantifiable evidence has led to firms investing in ML to digest multiple data sources and hone in on what is actionable intelligence vs. what should be disregarded.
Optimal use of technology will empower the trader to make more informed decisions based on quantifiable facts whether it is broker/venue selection or what price to trade at. But this also depends on ensuring diversity of thought within trading teams to optimise the efficiency gains. Increasingly, firms are recognising the value in creating freedom of thought in inclusive environments through race, gender and social diversity in both investment and execution decisions. Optimal use of AI and ML still depends on a strong partnership with human ideas and hypothesis if the industry is to move from reaggregation of existing liquidity to truly predictive correlation techniques.
A dose of optimism
Politics and economic policies will increasingly shape the future direction of the capital markets. Despite the US trade war with China, Brexit, upcoming European elections and the overall rise of protectionism across the world, the financial markets continue to demonstrate their ability to withstand turbulent times as well as their capacity to foster innovation.
The challenge for the industry now is the need to demonstrate true accountability and value for money. Either through insourcing IP or partnering to outsource, investment in technology and innovation will lead to further change in the European market eco-structure as firms struggle to keep on top of regulatory and cost pressures.
The challenge for the regulators is how to effectively regulate in this changing landscape; what the EU represents today and what it will be post Brexit; as well as how the EU will interact with third countries in an increasingly global trading environment. As noted on one panel at TradeTech, the risk is if the regulators misjudge, the market could respond with unnecessary innovation which may damage the fledgling European Capital Markets Union.
These challenges are set to ramp up as the world enters its fourth industrial revolution. Despite the rise in regulation and trade barriers, the world is becoming more globally connected. Future technology developments as well as the democratisation of access to data will help investment firms to become more efficient, which in turns will lead to improved efficiencies and better cost control. New job opportunities will be created as firms look to provide goods and services at more affordable prices. As the industry transforms, so will investment strategies and those who provide access to the investment network. Finally, the TradeTech debate is moving on from MiFID II, SIs and periodic auctions into the innovative era of data, technology and AI.
Machine learning models systematically scan newly arriving, anonymized data to identify anomalies.
The Cobalt programme launched in 2018 to help fintechs collaborate with the asset manager.
Users with different skill levels will be able to undertake machine learning and advanced analytics.
Fintech CEO discusses how AI applications have fared in highly volatile fixed income markets.
Exabel provides artificial intelligence and machine learning as a platform to non-technical users.