Stéphane Marie-François: Exploring transaction cost analysis

Senior vice president and multi-asset trader at Unigestion, Stéphane Marie-François, dives into the world of cross-asset transaction cost analysis (TCA) with The TRADE to discuss liquidity, automation and remaining limitations.

How can you leverage TCA innovation to generate better efficiencies across asset classes and markets?

Equity TCA has been existing globally for a number of years and is therefore much more matured compared to TCA on any other asset classes such as FI, FX and listed derivatives.

As of today, the lack of accurate data and the need of standards is particularly pronounced for fixed income transactions and over the counter (OTC) markets in general. Intraday data may not be available in many derivative or fixed income instruments making the use of the TCA not relevant. The request for quote (RFQ) workflow on FX justifies a limited use of the TCA as most of the costs are standardised across the brokers. Nevertheless, the fact that different asset classes can have similarities in terms of liquidity profile or regulation justifies the use of the TCA. There remains a huge amount of future potential for TCA in general once more clean data is available at a decent cost. Equity TCA is the most advanced and can be taken as a reference to improve and strengthen the other assets classes, we just need to adapt the metrics for each asset classes.

How can TCA allow traders to better interact with and visualise liquidity across asset classes?

TCA has become one of the most important components for buy-side trading desks in recent years. It gives the trader a quantitative measure and a deep analysis of exactly what has happened with each and every execution. TCA gives traders insights to better assess the difficulty of their trades in pre-trade, during the execution with real-time fills, and a deep overview in post-trade. There is a big topic at the moment about real-time TCA. I think it is at a too early stage because there is yet to be a strong solution. We have tools with some providers already but they are limited in terms of scope and options. I think we can do better and I trust the brokers and execution management systems (EMS) providers to deliver in the coming months. Brokers already have tools including machine learning (ML) or artificial intelligence (AI) which feed their SOR to improve trading decisions or behaviours of their algorithms. If we could have those kinds of tools available it would help the liquidity search and avoid toxic venues, while also readjusting size to avoid adverse selection etc. We could get the information in dedicated dashboards and/or alerts directly integrated into our EMS.

Do you think institutional interest in trading cryptocurrencies and digital assets is ramping up and how do you expect TCA to be evolved to reflect this?

Definitely, the appetite is growing. The high returns generated in the past and the endless innovation around the blockchain technology and decentralised finance (DeFi) are two sources of expectations. This asset class can bring diversification and can generate alpha. The limitations today are around the legal framework but we are making progress. TCA is a field that can be complicated because it relies on clean data. The cryptocurrencies world is still not that regulated and the prices can be different from one exchange to another. I think it is still at an early stage. Rebuilding a tape will be tricky. At the moment, taking an exposure to these assets is quite limited for our industry with the use of products like exchange traded funds (ETFs) which are expensive at the moment. I’m not sure if TCA would be a great help.

What limitations remain for multi-asset TCA, how can these be improved?

Certain asset classes such as fixed income are lagging in terms of TCA, and there is not a one size fits all. A good TCA relies on good (and public) data. Some tools just rely on the client’s data and it can mean the sample is not a good proxy of the market if the set of data is too small. Collecting data has a cost and companies will probably need to create a team to analyse it to justify the cost of spending. I think that as of today, most of the trading desks – except the big ones – use the tools included sometimes in the EMS (like we have with Virtu for equities and FXall for FX). To have clean, cheaper and consolidated data would be a game changer.

What role does automation play on a multi-asset desk, is full automation desirable?

As a multi-asset trading desk, automation is the key to success to strengthen order workflows. It helps traders to save some time so they can focus on more complex orders or code and develop other tools. Full automation depends on your style, your flows and your size. We went for partial automation as we think full automation may not be appropriate given our order flow. From our point of view and given the complexity of some of the orders we trade, there is little room for a full automation process which could incur any additional operational risk and unsolicited costs. We saw that during Covid where we decided to handle 100% of the orders by ourselves regarding the level of volatility in the markets. For larger companies it definitely makes sense especially when orders are liquid and small in dollar value.