A Spotlight on Automation and Algos with T. Rowe Price

How did you start your automation journey?

TB: T. Rowe have been looking at automation for a number of years. We purposefully took a slow and methodical approach to automation to ensure we did not go too big too soon and end up with disappointing results. When we started back in 2015, we initially piggybacked on the advice and relationships our equity colleagues had developed, looking at the automation processes they already had in place, and reviewing where FX can fall into scope. We started with spot transactions, G5 and major liquid currency pairs which had TCA attached to them meaning we could effectively review the quality of execution vs. voice.

We have learned that the critical piece to enable automation is robust analytics so you can undertake effective execution analysis. For anyone starting now, I would advise to start small with liquid G3 currencies, review the results and build from there. An incremental approach is key; you don’t want to go from a point of zero to automating a large portion of your flow, it needs to be done on a step-by-step basis.

BM: It is essential to know your order flow, for example knowing where you want to automate flow vs. where it is not suitable, and from there systematically analysing your order flow and developing rule sets to suit your auto-ex strategy accordingly. This is how you will get the most bang for your buck!

The name of the game has been scalability and efficiency. Over the last 5 years, our average daily ticket volume has increased by 50%, and in this time our automation has scaled up in lockstep meaning our teams have been supported to handle these increased flows. Without automation, we wouldn’t have been able to manage this increased volume. Not only have we been able to handle this volume, but we have also been much more efficient and cost effective without the need to increase headcount. 50% of our tickets are now automated, this is less than 5% of our dollar nominal traded. These are small tickets where traders do not add any value. Automation can give you time, and time for me is the most valuable commodity.

TB: Giving traders time to analyse markets, provide market colour and to generate more tickets for portfolios is where our traders and their critical skill sets are headed. Therefore, we’ve had to find a way to reduce time spent on the less value-add areas of execution and a way to automate this. This has given our traders the time and ability to not just execute but add significant value to the investment process. This has been particularly essential in negotiating volatile markets over the past year. In a world where traders are expected to have a multi-asset approach, having the ability to look at other instruments cannot happen if you have a blotter that needs constant attention; freeing up more time as a commodity is such a value add when it comes to automation.

What are your top tips for buy side when looking to select an FX EMS?

TB: Kick all the tyres and don’t be afraid of test driving an FX EMS! It is essential to look under the hood, understand the product, and establish the good, the bad, and the indifferent. Technology is moving so fast so the ability to trade differently with different instruments is moving on too. Look at forwards, swaps, NDFs, options and other derivatives; you still want to be able to execute these on the same platform and in the same pattern as spot FX. Unfortunately, not all platforms can offer this service yet so you must set your roadmap, know what you want out of the platform, and ensure that platform has the ability to add a new process to their product in the future that can work within your own timelines. There are three steps to onboarding a good FX EMS.

BM: Firstly, it’s important to fully understand the value proposition as well as the gaps an EMS can fill for you. Ensure you have clear and concise criteria to know what a successful outcome will look like. Secondly, you need to fully understand gaps you’re trying to fill and have a comprehensive inventory of the requirements you have, meaning you can lead the onboarding process rather than the vendor and have a good understanding of which workflows you want to have on the platform. Finally, having systematic and unbiased criteria to select the new EMS, which includes proof of concept testing and not just getting swayed by providers and sales talk, is essential. You need to know that the capabilities showcased will be realised from testing through to implementation.

How has increased automation on the desk changed your role? Have you found the experience to be positive or negative?

TB: Traders can be nervous about automation but from my use case experience, we wouldn’t be able to cope efficiently without it. The role of the trader is moving away from just execution and traders are becoming more of a focal point of the investment process, no more so than here at T. Rowe Price where we encourage this. We want to give the trader the time to come up with trade ideas and add to the portfolio.” Traders will always have a role as some aspects of FX are not suited to an automated process, you can’t put everything in the same bucket and expect the same results to come out. The trader will sit as the expert, watch all the automation that happens, and ensure nothing goes wrong while taking charge of the higher value trades and stepping in on automation when it is not delivering the results you need.

What benefits have you reaped from utilising FX algos?

BM: First and foremost, we have found the ability to capture spreads and achieve cost savings vs. risk transfer has been a real value-add. The benefits are numerous beyond this, such as being able to access internal bank liquidity at size; scalability and having the ability to multi-task while an algo is in flight; flexibility by being able to tailor strategies to the underlying execution objectives and reflecting market conditions you are working in; and finally better and more transparent TCA such as understanding the underlying liquidity and where it comes from. Taking full control is important. Not all algos will have a great outcome but it is about the ability to analyse the good and bad trades (i.e. negative outcomes) and learning from this.

TB: We have meetings on TCA performance, and we’ve found that in some currency pairs that are notoriously difficult to execute as a voice trade, we’ve discovered that algos offer better performance, whereas some other trades with tighter spreads it is difficult for a trader to outperform risk transfer price at times. Spreads change on an hourly and daily basis so it’s about having the relevant TCA that can give you this information, so you’re not surprised by an outcome. Ultimately, the benefits of an algo are the trader taking full control, working the market, and learning from positive and negative outcomes to continue to evolve your trading strategy going forward.

Is your FX algo usage increasing, and why?

TB: We initially increased our algo usage at the start of 2023, but as the year has gone on, we’ve actually decreased who we work with based on the results of our analytics. The analytics have shown us that we’re getting good results with certain voice trades rather than an algo, so we’ve reduced our usage in these areas. In other areas such as G4-G7 we’ve discovered that algos do lead to better execution. It’s about refining your algo strategy to match the environment you are working in.

BM: The bottom line is that algo usage increases in the right market conditions, and when these conditions change, we may reduce our algo usage to reflect market conditions. There are catalysts for increased usage. Firstly, launching the ability to split orders gives traders the flexibility to avoid tail risk. Secondly, we’ve done a lot of work on the analytics side, and as this gets more advanced with the adoption of new analytics providers and new data sets, this acts as an enabler for us to further understand the true areas where we can target algo use.

The average number of algo providers used amongst our FX community is 7. What is your algo selection rationale and what do you think is the optimum number of algo partners to work with?

BM: We keep our algo pool relatively tight, so our bank panel is purposefully small. We look to work with algo providers and tailor their strategies and their strengths to our trading activities, leveraging them in the right situation based on their offerings.

TB: Diversification is key. You would not operate with only 2 banks to execute all of your swaps or forwards business. It is essential that you try out and evaluate many providers to fully realise which institutions out there have better outcomes than others for you. In terms of onboarding, 5-7 is a sensible number, as long as you’ve done the right due diligence beforehand to get the right outcomes.

How can the buy and sell side work together to overcome integration, solution differentiation and transparency issues with algos?

BM: Both the buy side and sell side can take a more active role when working with vendors on integration. It’s on the buy side to understand their own requirements clearly and communicate these to both vendors and the sell side. For the sell side, it’s important to have technology that is scalable and portable across different vendors with features that can plug into multiple EMS vendors.

Providers should focus on strengths and continue to innovate to show their differentiation. It is not a one size fits all and it is not about one provider working for every single firm, they need to be homing in on their key VP and be able to adopt to a changing market structure- this is what sets vendors apart. Finally, vendors have come forward to launch unique data sets; if providers contribute to these data pools it will lead to greater transparency across the industry.

TB: The buzzword of the year has been interoperability and getting machines to talk to each other using the same language. It makes complete sense to enable interoperable set ups and seamless tech integration if efficiency is gained and costs are saved! For buyside and sell side to start on a process with algos, they may have teething issues, but they should pause, analyse and recalibrate as these issues can usually be resolved. Sharing experiences is how both the buy and sell side can learn from each other. It’s key to be able to adapt and evolve on the fly.