Consistent execution quality, reduced bias and greater resources to handle complex trades are all advantages that KBC Asset Management has seen from its automation of government bond trading. The next step will be to use machine learning to assess counterparties across fixed income instruments.
Increasing the automation of bond trading has clear advantages for an asset manager’s investors, but buy-side firms need to assess whether to buy or build tools, and how the role of traders will change as a result. Joeri Wouters, senior fixed income trader at KBC AM and Charlie Campbell-Johnston, managing director for AiEX and workflow solutions at Tradeweb discuss the lessons they have learned from their work together, including fully automated rates trading, supporting enhanced trade automation through partnership and the integration of AI into the trading workflow.
Dan Barnes Welcome to TV, I’m Dan Barnes. Joining me today are Charlie Campbell-Johnson at Tradeweb, and Joeri Wouters, senior fixed income trader at KBC Asset Management, to discuss the role of automation and improving, trading and delivering better returns for investors. Charlie, Joeri, welcome to Trader TV.
Charlie Campbell-Johnston Thank you very much for having me.
Joeri Wouters Glad to be here.
Dan Barnes So, Joeri, why should an investor care about trading automation?
Joeri Wouters The most obvious reason, of course, is efficiency, because we want to free up the time for the dealers to work on the more difficult high touch trades, or to work on projects that add value to our desk. Automation also decreases the risk of slippage. So definitely when we have to execute a larger number of trades at the given time in the day; fx the orders that we do for our passive funds at the close. Additionally, using a quant-based methodology for your counterparty selection, strengthens your best execution policy and improves it. It creates a uniform process across the desk for all the dealers, basically eliminating human bias. And finally, the same tools and analytics that we use to automate part of our flow, can also be used to improve our price discovery and counterparty selection methodology for high touch orders.
Charlie Campbell-Johnston What we’re seeing is not only that the capacity can increase as people trade more often, more frequently in more asset classes, with often no increase in headcount or certainly operational budget. But we’re also seeing examples of where automation is actually becoming an enabler to new types of trading activity. One example would be in the leverage space for derivatives, where you’re seeing automation bringing in the systematic funds to trade in a way that they wouldn’t have done historically.
Dan Barnes Which parts of the trade can be automated? And how does that differ by instrument?
Charlie Campbell-Johnston Typically, we will start with a client who is taking a sort of holistic view of their trading activity across multiple asset classes. And then over a period of time, as they get more comfortable and build up enough of a data set to be able to understand and quantify the results, then we can move further. And then also pull back if liquidity situations determine that now is not the best time to be automating. And that tends to be the same dynamic that we follow with a number of clients. But they tend not to have the same trading objectives at all. They differ quite significantly.
Joeri Wouters We started with automating a government bond flow already a few years ago. Initially, it took us hours to commend our selection criteria on a frequent basis. But with Charlie and his team, we created the workflow that only requires a few minutes now. The next step for us, and I think this one is an important one, is an internal project that we are working on at the moment. So, we are going to try to use machine-learning techniques to make predictions on the best counterparties to inquire. And we’re going to do that not only for the markets, but there’s a lot of data like COVIS, but we’re also going to try that for credit and EM. We’re going to use a broad set of data; like historical hits and pass rates, slippage versus green prices axes and so forth. I think it’s also very important to mention that this is new for our desk. There is no guarantee for success, but we hope to validate the first results by the end of the year.
Dan Barnes Does the best execution process change for low touch trades?
Joeri Wouters Yes, it changes in the sense that you will end up with a more uniform process across your desk. The increased use of data will result in a selection methodology that is less biased per dealer. That’s already very important. You also have to spend more time and more focus on the explainability of your selection criteria, because that’s often forgotten. You have to be able to understand and clearly explain to all the stakeholders why the tool took a certain decision. Additionally, low touch trades, we feel, do require an even closer monitoring as this is often rule-based. Constantly analyzing the data and the rules to improve the selection criteria is important. I think data management is really key here. And then finally, I think it’s still important that your dealers keep close track of the market and how the market is behaving so they can intervene and trade manually if needs be in times of stress, because as we’ve seen the last few weeks, the markets can turn very ugly, very quickly.
Since MiFID II there was a change in shift from the line by line execution, best execution policies to a more of a process-based execution policy, which actually sort of dovetails very nicely with the sort of rules-based approach. Because what you can do is you can implement a set of rules for a certain sector and evaluate the performance based on the criteria that you set that are within your policy. Joeri’s talked about slippage fx, but with different profiles, so some of the big index houses will care far more about matching a benchmark, or getting the trades done over a short period of time. So the ability to be able to say, ‘this set of rules has this outcome according to this metrics’, and analyze those systematically and adjust them according to different market conditions, becomes very valuable, but also changes, I guess, the nature of the role of the trader in some instances, to become much more data focused in analyzing that data.
Dan Barnes Charlie, Joeri, thank you very much.
Charlie Campbell-Johnston Thank you.
Joeri Wouters Thank you.
Dan Barnes I’d like to thank Charlie Campbell-Johnston and Joeri Wouters for their expertize on today’s show. Join us at TraderTV.NET and ETFTV.NET to catch our shows on other instrument classes and other investment types.
