What are the main drivers behind the development of your FX algo suite?
As a regional bank, our priority is to provide a boutique algo suite that suits our clients. For us this focusses our attention on Australia and New Zealand, so we offer our algos in AUD and NZD crosses as well as a USD base to best reflect our client needs. We want to only provide strategies that are relevant to our customers, so we stick with our strengths and offer a refined set of options that highlight where we think we can most add value to our client’s execution needs.
We are constantly measuring the effectiveness of our algorithm design and the execution logic, because we use the same stack for our own execution. Recently, for example, we identified an opportunity to change the positioning of our market-making orders on ECNs so as to improve the execution quality significantly.
In Westpac we have also been successfully using systematic trading methods for over 10 years, which has included the use of machine learning techniques. These techniques are used in pricing, execution and risk management algorithms, and optimise algorithmic logic and parameters based on pre- and post-trading analytics.
What are the key features and functionality available to your algo clients?
The ability to amend the aggressiveness setting of our liquidity seeker algo mid-flight is key, giving clients more control in how passively they want to execute. By being opportunistic based on available liquidity in the market, the passive settings of Seeker by design minimize market impact. Westpac’s algos by default also utilize a synthetic crossing capability, whereby when executing a cross currency pair, the algo will look at both the directly quoted liquidity and liquidity against the USD or EUR for each of our supported currency pairs and choose the best prevailing market price for each individual transaction.
How do you ensure your associated TCA toolsets meet the needs of algo users?
What is important to us is that our clients get from our TCA the information that is relevant to them. TCA design is therefore very collaborative and we are always looking to enhance our TCA based on feedback. Clients look at a variety of metrics to serve as comparisons to the algo executed rate and we want to meet them with what best expresses their execution metrics. We’re currently working on our next release of TCA enhancements as well as making it easier for our clients to send their data to third parties for external analysis.
We’re currently working on our next release of TCA enhancements as well as making it easier for our clients to send their data to third parties for external analysis
What should clients be looking for in terms of the liquidity which FX algos can access?
Sometimes we see clients very focused on a particular metric – for eg spread minimisation or maximum internalization – without considering whether this is actually important for their overall execution aim. We see our role as assisting customers in articulating what is really important to them, so we can help them choose the best product suited for their needs.
We want to add our own experience as an execution desk to the value proposition of our algos. For example, we have noticed that a focus on completely passive (i.e. spread minimisation) strategies may lead to adverse selection and not actually benefit the client. Therefore, we choose not to offer passive only algos and instead prefer to offer an algo like Seeker that balances the benefits of being passive (to minimize spread crossing and market impact) with the benefits of being active (maximize fill ratios and minimize adverse selection as well as opportunity cost).
Our own experience in liquidity selection is also important in the product design. As bank execution desk we are constantly testing and evaluating different liquidity sources and execution techniques, and this knowledge is a valuable element of the algo the client selects.
Do you believe that the way FX algos performed over the past year has had a lasting impact on rates of adoption or the types of algos clients are interested in?
Yes. With market volatility at the beginning of the pandemic there appeared to be a strong increase in volume, either through existing clients putting through more orders or new participants experimenting with the product as spreads widened. Since then the popularity has waned thanks to incredibly tight two way eFX pricing but algo use still remains elevated over pre-pandemic levels, with some of the new users remaining.
For our own Australia and New Zealand based clients, whose business days start at the end of the New York session, algos represent a useful tool in their arsenal to consider when trying to clear larger amounts in markets that are noticeably lower in liquidity than the London time zone.
What impact will the recent changes to the FX Global Code have in terms of FX algo use?
With the recent three year refresh of the FX Global Code there has been a huge focus on standardising disclosures throughout the market. This is to ensure liquidity providers are making these disclosures clear, accessible, transparent and consistent across the street so it is easier for liquidity consumers to make informed decisions. There has also been increased focus on last look and whether there is a need for hold times in the market as well as ensuring liquidity consumers can access the information required to understand why their trades may have been rejected.
We have been using systematic trading methods for over 10 years,
which has included the use of machine learning techniques
What will be your main areas of focus going forward?
Our focus will remain on what our clients want and need. In the TCA space, we expect to see clients continue to be more specialized in what they look for in TCA – more customization requests and likely an emergence of more third party TCA offerings. At the same time that needs to be balanced with the direction we are seeing from the FX Global Code which is towards standardization and transparency, particularly in disclosures. We will continue to focus on R&D to optimise the balance between passive and active execution and refine the pricing of internal liquidity to maximise internalization.
