Michael, we last interviewed Shell in FXAlgoNews 6 years ago. At that time the firm was a pioneer in adopting algorithmic FX trading. However, pioneers sometimes “catch the arrows” so what advice would you give to other firms who are just starting out on a similar journey?
Whilst it may sound obvious, the most important element of algorithmic execution is actually quite simple, understanding your objectives. Only once you understood what you want to achieve in terms of your execution strategy is it possible to determine which algo strategies are appropriate. After that, it is all about data. Once you have set your strategy, the data is a key element to understand how well the specific products, which at times can sound homogenous, delivered against that strategy. The subsequent selection of products, and associated settings available on each of them, can subsequently be tweaked, as can the broader strategy and objectives.
What are you trying to achieve by using algorithmic FX trading and what are the most important benefits that it delivers for your team?
There are a number of potential benefits that we look at; liquidity and smart order routing, automation and confidentiality. Liquidity and smart routing really depend on the intelligence of the algo, and its ability to clear risk across different venues in a way that does not cause market impact.
Automation allows us to clear risk on a no touch basis, routing trades automatically to an algo to clear time sensitive positions in an effective manner. Confidentiality applies mainly to positions over a certain threshold, to limit the risk of information leakage. Whilst these benefits are all tangible, there is a clear trade off against the cost of the algo. These products aren’t free, algos are therefore only one of the tools that we use to manage risk.
How do you go about measuring the effectiveness of your algorithmic FX trading?
Data is absolutely key to measuring success, I don’t think it is possible to truly understand the success or otherwise of an algo strategy (vs RFQ/RFS/voice) without it. When we first started in this area, we built our own internal TCA tool, which allowed us to assess algo strategies against each other and against our other strategies. In recent years, as third-party TCA tools have evolved, we have discontinued to this, and now use BestX to assess performance. This includes all FX execution strategies, not just algos.
As the use of execution algorithms grows in popularity and becomes more prevalent in FX do you think there needs to be more regulatory oversight of this form of trading and what role can large firms like Shell play in helping the industry find the right balance?
Governance in the algo, and electronic execution space in general is absolutely key. This is vital for a couple of reasons; to ensure markets continue to operate in a fair and effective manner, to ensure that clients achieve their execution aims, whilst having confidence in algorithmic and electronic execution strategies. Whilst it should go without saying that the algo providers have a huge responsibility to deliver this, I also believe the buy side have a responsibility to understand the products that they are using and the way they are using them. Whilst this may not extend to in-depth market micro structure discussions, the buy side should seek to actively challenge their providers to understand how the product works and particularly the control and governance mechanisms that form part of the algo (kill switches for example). Shell is part of, and welcomes, the progress in this area as part of the 3-year review of the FX Global Code.