Viewpoint: Realising the benefits of TCA

Mark Ford, managing director at ISS LiquidMetrix, spoke to Best Execution about the evolution of transaction cost analysis (TCA).

How have clients changed the way they use TCA? 

We have observed a sea change in the way clients now view TCA. It is no longer seen as an optional tool or tick box exercise but has become an invaluable and mainstream workflow tool for traders. It is increasingly being used across the whole trade lifecycle and has become much more a part of the investment process to generate better returns and optimise portfolios.

One reason is MiFID II, which has made asset managers more accountable, and as a result they are demanding more from their brokers. However, volatile markets have also been a driver. Equity returns have declined, squeezing profitability and margins, which in turn has made firms look for deeper insights into performance characteristics.

In general, TCA has always been about data and analytics but what they are collecting has changed. We are seeing demand for customisation and more granular information. For example, they want to know the types of venues – lit or dark – being traded on, or whether strategies are aggressive or passive. This is a knock-on effect from MiFID II, but we are also seeing the same trends in the US as in Europe.

Is there a different level of engagement with clients now?

The demand for greater analysis and data on orders and executions has meant that the simple reports of the past no longer suffice. They have proved limited in their ability to illustrate different trading scenarios. There is prevalence from both buy and sellside firms for more detailed interactive analytics through visualisations and displays used in best execution reviews by clients. We have found that this requires a closer integration with the client OMS/EMS (order management and execution systems). It extends far beyond the liquidity indicator tags now widespread after MiFID II became effective, but into the parameter settings for algos and logging of changes during the life of an order. This can be demonstrated through a change in the urgency of an algo as the trading performance will dramatically alter once this parameter is increased. As a result, each part of an order needs to be analysed separately in sub or child orders/placements to fully understand the impact of these changes.

What is happening currently in the field?

Currently, I would say it is like new wine in old bottles, and that in general the development of TCA has been more evolution that revolution. This is because measuring trading costs is one thing but interpreting them is another. There have been areas of investment such as cost estimation models and peer group benchmarks which can help traders better understand the costs of the trade. However, there are issues. For example, many TCA systems have cost estimate facilities, but these models have to be improved to reflect not just fragmented liquidity, but also to take account of the actual order book volatilities for the instruments. As for peer benchmark analysis, it is a performance measure that needs the threshold redefined. Simply stating that performance is within a quartile is not useful in either properly assessing performance or giving any indication of what can be improved. Instead, AI techniques such a machine learning, are more commonly used to provide better indications of what is good or bad performance.

We are also seeing an increased focus on risk and pre-trade costs analysis. It has been under the banner of TCA for several years but there are new, more sophisticated techniques that use market data to calculate the costs, identify patterns and behaviours. We are seeing increased use of TCA to determine which algos should be included or dropped from the algo wheels.

Although TCA is a global trend, what are the regional differences in the use of TCA?

There are differences. For example, the quality and granularity of the data are good in both North America and Europe, but they require different analytics to reflect their different market structures and regulation. In Europe, from MiFID II, we had the emergence of different venues such as systematic internalisers and periodic auction pools. The US, on the other hand, has far more fragmentation and a different microstructure with rebate exchanges (payments from the venue for orders), as well as the categorisation and performance of so called “grey markets”.

Asia Pacific tends to follow Europe but for some markets, we have to use wider tolerances on data latency for execution times. We do not use different algos but there is much less choice of trading venues with only Japan and Australia having market fragmentation.

What do you see as the likely trends in 2023?

As returns on trading equities become smaller and more volatile, there will be ever increasing pressure to reduce trading costs to maintain overall performance. This will continue the need for more in depth and granular information and analysis. I think we will also continue to see more attention paid to efficient ways of identifying good sources of liquidity in general as well as on a stock-by-stock basis to fine-tune to most effective trading strategy based on that instrument’s liquidity characteristics. We also forecast increased use of algos across the desk and not just by the head of trading

How do you expect TCA / best execution to change in the future?

I see several developments. First, there will not only be more data being produced but better quality and that TCA will no longer sit outside but be more integrated onto EMS/OMS with APIs between the systems and feedback loops. There is likely to be less reliance on a trader’s discretion because the data will drive the decision making, but this can only be achieved through greater integration.

I also expect de-facto reports to always be available, and potentially the standardisation of performance benchmarks. Use of AI techniques will continue to increase and be used to recognise behaviours and determine optimal algo selection. Combined cross-asset performance should also become popular instead of what we have now – currently each asset class has to be assessed separately.

Last but not least, I believe there will be a greater focus on order routing, which is the path the broker chooses to execute a trade. If the order is marketable, the broker looks at the trade characteristics, such as liquidity of the stock and other factors, to determine the path to achieve best execution. Increasingly, we think this type of analysis will be reviewed not only by brokers but by the buyside as well as part of their best execution process.

To learn more, contact

©Markets Media Europe 2022