Andreas Koenig – Talking algos with a leading FX practitioner

August 2023 in Buyside Interviews

Amundi is Europe’s largest asset manager by assets under management and ranks in the top 10 globally. The company manages over 1.476 trillion euros of assets across six main investment hubs. Utilising its unique research capabilities and the skills of close to 4,500 team members and market experts based in 37 countries, Amundi provides retail, institutional and corporate clients with innovative investment strategies and solutions tailored to their needs, targeted outcomes and risk profiles. Andreas Koenig is responsible for Global FX within this huge organisation and we asked him to tell us a little about how he views FX algorithmic trading and in what ways Amundi is leveraging these increasingly important and powerful toolsets.

Andreas you have worked in the investment industry for over 25 years. How did you get involved in FX and what do you like about working in this market?

I actually started in FX and have been in it ever since. It was partly by coincidence. I started a trainee programme in global markets after university, but during this time the derivative business was carved out into a derivative subsidiary and the programme was terminated. At this particular time, I was in the FX options team and stayed there. I like the truly global nature of this market, with its deep liquidity and flexibility. It is by definition a real absolute return asset class, where all focus lies on alpha, as the market itself has no underlying beta.

How does the FX unit in Amundi sit within the overall structure of the company and what are its key tasks?

The Global FX team sits under Absolute Return and is located within the Global Fixed Income team in London. It is a team of FX portfolio managers and the main task is the management of the dedicated FX funds and FX overlays. In addition, we are the point of contact for colleagues and clients internally and externally for questions concerning FX.With our expertise in FX markets and instruments, we support other PMs and teams. The trading or execution team is completely separate from the PMs, so the answers concerning execution and using of Algos are from the colleagues there.

We are only using FX algos for the execution of large orders or where liquidity is an issue

We are only using FX algos for the execution of large orders or where liquidity is an issue

Please tell us a little about what your job and day-to-day responsibilities usually involve

Our main task is to analyse and evaluate global FX markets with the goal to extract positive alpha out of the moves of currencies. Currencies are influenced by a large variety of factors. The basis of our daily job is to analyse, evaluate and rank these factors as well as following the economic and political developments in the respective countries.

The range of tasks extends from active position taking in currencies to create yield, to passively hedging currency risks. This completely depends upon the risk and return expectations and constraints of the respective investors. Due to the flexibility and deep liquidity of the FX market, we can offer tailor-made solutions for almost all of our clients’ needs. The responsibilities are mainly defined by return expectations and the connected risk management needs of the investors.

Within the FX team, we split responsibilities e.g. fundamental macro, quant, derivatives etc. This gives us the ability to specialise and focus.

Having said that, the co-ordination of tasks and keeping the overall development of the portfolios in mind are essential responsibilities as well.

The execution of our orders is done by the trading/execution team, which is a completely separate one

The execution of our orders is done by the trading/execution team, which is a completely separate one

Could you briefly outline the sort of processes that are involved with the execution cycle of a typical FX order you deal with. For example from initial order generation then onto compliance and risk management tools and through to the ultimate execution systems that your trading teams use.

The FX team is a team of PMs where investment decisions and idea creation take place. The execution of our orders is done by the trading/execution team, which is a completely separate team. The process is very straightforward. The idea creation is followed by the investment decision, and an order is created in the front office system. Compliance checks, risk constraints and client restrictions are checked in the system before the order is sent to the execution team.

The traders pick up the order and decide the best way to execute the order in terms of instrument, market impact, price, PM constraints and transparency.

After finishing execution, the trade is booked and processed through the mid and back office to the custodian. The PM can see the trade in the system and manage risk and exposure from there.

We are currently using new FX analytics to prove best execution to our clients and improve our general workflows

We are currently using new FX analytics to prove best execution to our clients and improve our general workflows

In what ways has MiFID II with its focus on transparency around best execution influenced how you go about measuring and evaluating your own trading activities and has it encouraged you to explore new execution alternatives and methods to undertake them?

As mentioned, we strictly separate Portfolio Management and Execution.  As a PM, the main focus is to find attractive opportunities in which to invest. In this respect, MiFID is not a significant influence. On the execution side, though MIFID II requires additional monitoring and reporting procedures, all market participants need to follow sufficient steps to achieve best execution. We fully comply with these requirements.

Some firms feel it’s important to use a stable and mature platform rather than the latest technology. What’s your opinion on that and how much importance does Amundi place on continuously working to improve its own trading infrastructure?

Using stable platforms is very important for Amundi. The onboarding of all the portfolios we manage takes a lot of time, and we need to be sure of the reliability of our trading infrastructure. That being said, there is always a need to ensure we are innovating to reflect market changes.

When using algos, we are trying to get a better price than what we would achieve with a risk transfer price by working the order and carrying the execution risk

When using algos, we are trying to get a better price than what we would achieve with a risk transfer price by working the order and carrying the execution risk

How long have your trading teams been utilising algorithmic execution in order to handle FX orders and what factors usually determine the type of algos they use?

We have been using algorithms for FX execution for about 5 years. We aim to achieve a quick execution while minimizing market impact, so we consider the pool of liquidity the Algo can access and the aggressiveness of the strategy when determining which type of Algo to use.

Many large asset managers choose to utilise bank-generated algos as they believe it gives them better access to different liquidity streams and trading venues. How does Amundi source its algos and what are the differentiators when choosing one provider over another?

Amundi only use Algos provided by banks. Some of them allow their Algos to internalize flows and we think it is a great addition to the external pools of liquidity. The main differentiators between different providers are the different strategies used by their Algos, the flexibility in the parameters before and during execution and their capabilities of internalization.

What are the important things you are trying to achieve when using algorithmic FX trading techniques and do your traders still have the discretion to deviate from their default execution strategy when using algos?

When using algos, we are trying to get a better price than what we would achieve with a risk transfer price by working the order and carrying the execution risk. The execution method is at the discretion of the trader.

What types of FX algos (TWAP, VWAP etc) do you currently find most useful and what features and functionality are you looking for that will add the most value for your trading teams?
The two main strategies of algos we use are:

1. aggressive strategies that sweep the market up to a level and iceberg the rest of the order
2. passive strategies to limit market impact in illiquid currency pairs, with the option to become more aggressive when liquidity improves or the level becomes more attractive

How involved do your trading teams get in the testing and fine-tuning of the FX algos they are using?

They are constantly giving feedback to the algo providers on how their new and old algos perform and provide them with the functionalities and behaviours they would like to see in their algos.

Are you using FX algos for every trade or only for specific situations, for example executing a larger block trade or dealing with more illiquid currencies?

We are only using FX algos for the execution of large orders or where liquidity is an issue and a risk transfer price is prohibitive.

We frequently talk about the two schools of thought on whether to let FX algos do their job or to go for a more hybrid, human-algo approach. What’s your opinion on that?

We like to have the flexibility to choose whether we want to leave the algo do its job by itself or to give it very tight parameters that we would monitor and change during the execution, as different market conditions call for different approaches. Ultimately, human know how is very important and monitoring the machines is essential.

What do you see as the biggest challenges in trying to accurately compute execution metrics in FX?

For us, the biggest challenge is properly capturing the market depth and the risk of a transfer price on the full size of our order if we decide to work an order.

Granular TCA is important to ensure that algorithmic execution strategies genuinely achieve a better result than more traditional trading methods. How do you tackle this and what sort of benchmarks are useful?

Our most important benchmark is the risk transfer price at inception, as trading at a risk transfer price is usually the alternative when we receive the order. The second most important benchmark would be the arrival price, as it allows us to capture the true execution cost including market spread and market impact.

Do you expect demand for more complex FX trade analytics will increase and how would that fit with your own requirements at Amundi?

New FX analytics are becoming available on a regular basis and we are currently using them to prove best execution to our clients and improve our general workflows. We believe the demand for these analytics is growing as clients request more and more transparency.

There is not always a consistent correlation between the cost of one particular algo and the direct benefits it delivers. What would you like to see done to address that?

To address this issue, we are requesting that all of our algo providers be on the same fee structure. That way, only the performance of the algos matter in the choice of a strategy for the execution of an order.

Are you satisfied that leading FX algo providers have solved most of the potential threats and conflicts of interest that can sometimes arise when using these toolsets and what sort of due diligence does Amundi undertake with regard to this?

Before working with a new algo provider, Amundi ensures that proper Chinese walls are in place between the electronic and voice trading desks at the algo provider. With this in mind, we are confident that most providers have limited any potential threats.

Are there any specific things that you would like developers and providers of FX algos to put more focus on? For example by giving you better insight into the venues and liquidity providers that they route to and showing you performance data on these.

More transparency on the live market depth on different venues could definitely be helpful in order to help assess the current liquidity and the best strategy to use. Generally, we have found algo providers receptive to our feedback.

What do you think will drive increased use of algorithmic FX trading toolsets by buyside firms over the next few years and do you expect to see Amundi making more use of them?

We believe FX liquidity providers will find it increasingly challenging to provide liquidity for larger-sized tickets, and as such the use of algos to work large orders will become the norm. Additionally, the move towards greater transparency lends itself to the increased use of algos. Therefore, we think that the development will go in direction of electronic trading. Generally, this approach to trading and execution has an advantage and leads to better execution results. But there will also be occasions when volatility spikes, when liquidity deteriorates and electronic trading and algos may struggle.

As such, we believe that it is important to be well connected to counterparties, to provide human execution expertise and to rely on multiple market access in such an environment, in order to deliver best execution in all market situations.