The use of algorithmic trading is in its relative infancy in Australia, especially when it comes to FX. Australia-based financial research firm Peter Lee Associates has only been tracking algorithmic trading levels in its studies for the last three years and currently finds usage is largely limited to the large volume users – those with A$10 billion or above in annual FX trading volumes.
Among these large volume users, according to Peter Lee Associates’ annual FX study, the number of financial institutions respondents that cited algo trading increased from nine in 2019 to 12 in 2020. The figure dropped to 11 in 2021, although PLA managing director Cameron Peter puts this down to sample changes in the survey rather than a fall in adoption rates. “The 11 users equates to one-third of the large volume users in Australia using algorithmic trading,” says Peter.
It is not just the number of users that is modest but also the percentage of total trading done via algos. When asked what percentage of total volume is executed via algorithmic means, six users reported 3% or less, one user reported 5% while four users reported more than 10%.
When it comes to the type of algo trading employed, the majority (64%) use schedule-based algos such as TWAPs, while a smaller figure (45%) use other types, mainly icebergs and slices with some users employing both types.
The 2021 study was the first time that PLA tracked algo usage among corporates. The 2021 Corporate FX study showed a marked increase in FX trading volumes, up by 7% to A$323 billion, the highest reported volumes in 13 years, which Peter ascribes to a sharp lift in resource prices, particularly iron ore.
So it was not surprising that algo usage, as with the financial institutions and other buy-side traders, was limited to the large volume users, those with A$1 billion in annual FX volumes or more. Among those large volume corporate users, five reported using algos. Of these five, four were in the natural resources sector.
The other buy-side sector where FX algo use is likely to increase is the super funds market. In the last year, Australia’s A$3.3 trillion superannuation sector has seen a continued wave of consolidation, encouraged by the country’s regulator. There were 15 mergers in the sector in the 12 months up to October 2021, which is driving the market’s structure to a handful of mega-funds, or super funds. And underpinning this all is a vast pool of assets that has been amassed due to Australia’s compulsory pension saving rules.
In an FX sense, these super funds will likely turn increasingly to algo trading in order to minimise the market impact of their trades in what remains a relatively illiquid market in Australia.
So how are Australia’s leading banks developing their FX algo offerings to cater for these changing market developments? There are numerous drivers for the increased adoption of algorithmic FX trading across Australia, says Paul Scott, Head of eFICC Capability at ANZ. One is the best execution requirements that have come from regulatory initiatives such as MiFID II and the Global FX code of conduct.
More clients are also using algo orders rather than ‘at best’ orders to reduce their market impact as well as information leakage. Greater access to transaction cost analysis (TCA) has helped in this regard.
But as the adoption of electronic trading continues to increase throughout the FX market, the buy-side’s understanding and expectations regarding connectivity, venue selection, internalisation rates, global infrastructure and support becomes increasingly important, says Scott.
While Australia has a large and diverse community of buy-side players, it is essentially binary when it comes to the use of FX algos, says Scott. On the one side, you have the large institutions and super funds, which have become sophisticated users of algos and are demanding a wider range of trading tools. They are also demanding greater control over their algos. An example includes using pause-amend-resume and dynamic functionality over the algo execution logic.
But away from the super funds, the use of algos is much more basic, as are the demands. Very few fund managers or corporates are using any algo more advanced than a TWAP or limit type order algo. This bifurcation of Australia’s buy-side traders necessitates flexibility in bank algo offerings in terms of venue selection, algo suite and TCA, either through a ‘client self-serve’ or full-support model.
“The large real money/superannuation firms are growing exponentially and as a result the need to invest offshore necessitates the importance of FX overlay and execution to protect offshore earnings,” says Scott. “These flows can be significant in size and may require benchmarking which are two of the traditional reasons for using FX algos. The Australian market can be less liquid at times than other markets, so market impact is of key importance. That is why ANZ has focused so much on liquidity management and reducing market impact.”
But as the rest of the buy-side clients increase their use of e-trading solutions, we will see further focus on streamlining straight-through-processing (STP) given the more complex nature of post trade activities for these sectors, says Scott.
“Apart from requiring allocation, netting and block trade processes, the buy-side and more broadly the FX market are looking at holistic credit, margining and clearing solutions. Equities has always been leading the charge initially in electronic trading and then client Algo solutions, as eFX trading become more broadly accepted we expect FX algos to follow a similar pattern of adoption,” adds Scott.
In a market where TWAPs are still the primary product, how do the banks provide any differentiation to their client when it comes to client algos? “The real differentiation between the different bank algos is around access to liquidity, interaction with that liquidity and understanding the impact of that liquidity” says Joel Marsden, Head of FX Algo and Order Management at ANZ. “Liquid management is key.”
“It is important to acknowledge exchanged-based trading concepts do not necessarily correlate to best execution in a fragmented global OTC FX market. The definition of what is deemed ‘best execution’ may vary greatly from one counterparty to the next due to differing trading expectation outcomes such as benchmarking, rejection/ fill ratios or market signally and implementation shortfall,” says Scott.
ANZ has recently embarked on a recoding and updating of all the technology, infrastructure and analytics behind its suite of client algos which has now been migrated across to the new platform. This ensures the bank can respond to changes in market dynamics, client behaviour and regulatory adherence. “At the heart of its offering is a robust but flexible order and liquidity management framework which allows scalability for their algo suite,” says Marsden.
“We are always looking to roll out new features to our algo toolset through a regional customer driven lens. We have just launched a dynamic algo to the multi-bank portals that tracks the market according to volatility and liquidity conditions based on customer requests. Forward settlement dates are important to our Australian institutional customers as algos become more entrenched with treasury management. We are currently looking at how to best support that seamlessly for our customers,” says Marsden. A limit order/pegging order algo is set to be added in the next few months.
The reason for the expansion of the FX algo suite is the growing size and sophistication of the Australian client base who demand a broader range of execution options and the ability to control their own trading, says Scott.
“Another key value proposition for our Australian clients is the way me manage cross-currency orders which aligns with ANZ Institutional bank’s value proposition of being the best bank for clients driven by regional trade and capital flows. Our algorithms look to optimise trading through both the underlying currency pairs, in conjunction with the tradeable cross itself, to better access market liquidity and improve customer execution,” says Marsden.
Sustainable liquidity is critical for ANZ’s client algo offering as well as their broader FX business. “We have developed advanced trading analytics that ensure we have a granular understanding of the impact from our liquidity providers as well as our own trading behaviour on the market,” says Scott.
“ANZ has leveraged our internal liquidity aggregation product and smart order routing techniques to develop a range of algos that can access a diverse range of liquidity venues,” says Marsden “Venue categorisation; black listing; Full amount logic, skew exclusion logic; speed bumps and real time trade analytics are all key elements of our liquidity management. We also produce monthly “Liquidity provider” reports that allow transparency and two-way communication on constant improvements to sustainable liquidity.”
Scott says that ANZ’s entire eFX stack is internally developed with no external vendors and is hosted by ANZ and this allows the bank to respond to changes in market dynamics from a geographic or client perspective.
“There are four issues that come to mind with internally-built solutions. The budget and resources required to build a globally distributed eFX solution is significant. Credit and access to liquidity is fragmented within OTC FX. Compliance and regulatory standards are increasingly a barrier to entry for the algos market. And benchmarking against peers can preclude the use of a proprietary platform.” Given the above, there is still little movement among most buy-side firms to develop and build their own FX algos, says Scott.
Regulation is certainly a huge driver in the adoption of algos, as each individual fill can be analysed both on a pre and post execution basis, says Marsden. There has also been an increase in the availability of third party TCA analysis which has not only helped firms meet their regulatory requirements but also encouraged greater use of algos.
Trade processes have been streamlined in order to drive down costs and ensure compliance with ongoing regulation changes, says Scott. “We monitor and police our unique liquidity venues to ensure there is no excessive hold times, market impact or rejects and as an institution encourage all market participants and venues to sign up to the FX Global Code of Conduct.”
“As part of regulation and client disclosure documentation requirements, it is essential to ensure the client understand the algos they are implementing, therefore the development effort is placed in liquidity access, scalability, analytics, STP and overall support versus increased complexity, says Marsden.
Initially, the leading algo providers emphasised the importance on “internalisation” rates of an algo, however this has shifted with the focus on market signalling and subsequent benchmarks to arrival price, implementation shortfall, child order spread performance, risk transfer.
ANZ offers the choice of including its internal liquidity or matching only with external venues and providers, says Marsden. “It is the sole discretion of the client. By utilising ANZ’s liquidity, we are able to show unique liquidity and advise clients on local liquidity conditions, especially during the Asia time zone given the breadth of our branch network and franchise footprint.”
In general, there is more responsibility on the clients to ensure they are responsible for understanding their execution, this has seen a move away from more complex ‘black box’ algos, says Scott.
“In response to global regulation, such as the FCA’s algorithmic trading compliance framework, the FX global code – which ANZ attested to in December 2017 – as well as the recently developed ‘Algo Due Diligence Template’ – we are seeing a drive towards greater transparency, fairness and consistency in relation to FX execution,” says Scott.