The last two months have seen markets around the world gyrate as investors struggle to understand and react to the global COVID-19 pandemic. As traders are forced to quickly examine strategies and change tactics on the fly, understanding the implication of each trade has become more important than ever.
What’s more, as the buy-side grapples with higher costs, fee compression, volatile markets and regulation, many industry participants are turning to Artificial Intelligence (AI) and machine learning technology as they hunt for further efficiencies in their trading processes. Building these new processes in-house or outsourcing to technology providers has now become increasingly common.
AI and machine learning have been used by the sell-side for a long time to automate menial tasks performed by sales traders. Buy side participants are now starting to use these technologies in a variety of additional ways, such as; creating new processes for price, liquidity discovery and execution algos. AI and machine learning technologies have also been used to improve transaction cost analysis (TCA) at a time when asset managers are legally obliged to show regulators that “all sufficient steps” have been taken to achieve best execution.
According to a 2020 Greenwich Associates report, TCA is now used by 19 out of 20 firms in Europe and nearly 9 out of 10 in the United States. The Americas are continuing to catch up to Europeans in using TCA. This includes sophisticated investors. 10% of asset managers in the U.S. have a TCA specialist, whereas 25% in Europe have a dedicated TCA specialist. An outsourced TCA provider can be a great alternative to firms without a dedicated TCA specialist and can easily help an asset manager as well as the trader, and, even in some cases the risk team.
Already on the market are equity-specific TCA tools that enable the buy-side to analyse performance on a pre-trade and intra-trade basis. These technologically advanced tools take into account a myriad of real-time data points to help traders determine if they are achieving best execution for their orders by monitoring costs along the lifecycle of the trade.
TORA’s pre-trade equity TCA solution uses machine learning to examine the core attributes of an equities trade, including spread, volatility and volume consumption. This has the power to estimate market impact before the trader enters the market. An AlgoWheel or broker selection process in particular, have grabbed the attention of the buy-side. TORA’s AlgoWheel uses AI to help firms identify the best broker algorithm for an order via an easy to read analysis of both historical and real-time prices, and volume time series.
This combined with the properties of historical equity trade execution of orders with similar market capitalisation, sector and volume consumption has revolutionised transparency and the process of trading within the market. TCA providers, buy-side and sell-side research teams are all continuously looking at new ways equity TCA can be developed. For example, building pre-trade TCA and recommendations based around live market volatility and liquidity characteristics throughout the day.
With MiFID II extending best execution requirements into other asset classes and requiring more transparency and post-trade reporting in fixed income for the first time, there has been great interest in how AI and machine learning may be able to benefit other asset classes. Post-trade TCA for equities has been available since the 1990s and has reached a certain level of maturity, but the same cannot be said for FX and fixed income TCA.
For Non-equity, AI and TCA are increasingly being looked at from the perspective of transaction cost analysis rather than for the algorithms themselves. While increased regulatory scrutiny has created interest in using these technologies for TCA across all asset classes, according to Aite’s report there is still some way to go until its use case for alpha generation has been fully realized.
The pandemic and subsequent spike in market volatility was a chance for artificial intelligence to prove its value in a time of crisis. Traders with machine learning enhanced TCA were able to quickly understand the impact of widened pockets of more illiquidity and widening spreads – and adjust trading patterns in response. The benefits firms found in this crisis augur even greater rewards once advanced TCA is implemented across all tradable asset classes.