Westpac is both a consumer and a provider of execution algorithms. Matt Fitzpatrick, Head of eFX Pricing and Flow Trading models, explains how Westpac’s algo expertise has taken lessons learned from being the first Australian bank to design and operate a fully electronic trading desk, set up a segregated Benchmark trading operation and offer in-house developed execution algorithms to clients in 2016. From the start, Westpac made a conscious decision to provide a refined and targeted offering that utilises over a decade of eTrading expertise.
PRICE
When assessing algo performance, Fitzpatrick’s team thinks the most important question is: Does the execution algorithm achieve a great price?
For many clients, this may be difficult to answer with a reasonable level of confidence due to the statistical noise in markets. Fitzpatrick illustrates this with an example:
Suppose we have a client who wants to determine if Bank A or Bank B has a superior algo. For simplicity, we assume the client executes in independent and identically distributed market conditions with the same volume, currency, and accessible venues, that the algos from Bank A and Bank B have the same standard deviation of fill rate of 2 basis points and Bank B can achieve an average fill rate that is superior by 0.1 basis points. Even in these statistically convenient conditions, Fitzpatrick believes the client would have to run both algos about 3,000 times to have 95% confidence that the algo from Bank B is superior. If they run each algo once a day, that would take about 10 years to get to the answer!
DATA
Commonly, algo users are provided with data on comparable fill prices, pre-TCA and fill rates to assist with assessing algo performance. Unfortunately, Fitzpatrick’s view is that this data in isolation does not allow clients to truly analyse best price.
Having run a successful electronic market making business for over a decade Westpac has collected their own data and established a team of data scientists and quants who ensure its reliability. This enables Westpac to fine tune prices sent to clients and design high performing execution algorithms. Fitzpatrick explains that, as a market-making bank, there is ample opportunity to see thousands of trades each month in frequently traded instruments, allowing the team to work through the confounders and tease out statistical significance.
Westpac has focused on a targeted algo offering based on internal data and experience. Gin Devoy, Head of eSales at Westpac, tells us they made a decision at the outset to only make available solutions to their customers that utilise the internally proven execution tools available to Westpac’s own trading desk. Devoy says “We focus on minimising adverse selection in our own trading and strive to deliver the same for our clients. Our Liquidity Seeker Algo takes this into consideration, where we make use of our internal fill data to optimise the mix of order types. This is where we are balancing minimising market impact against the risk of adverse selection on passive fills.”
Fitzpatrick concludes, “Westpac’s offering provides clients with access to the benefit of our long-standing investment in eTrading expertise and the strength of our AUD and NZD franchise, which is something our clients appreciate. Therefore, our clients have an opportunity to internalise with our eFX Trading desk and can feel confident in our commitment to minimise market impact and maximise internalisation.”
This material is of a general nature only and is not intended to constitute a research report or reflect any recommendation or financial advice. It has been prepared without taking into account your objectives, financial situation or needs and investment decisions should not be based on it. Whilst every effort has been taken to ensure that the contents are correct and any opinions, conclusions are reasonably held based on information at the date of this interview, Westpac banking Corporation (ABN 33 007 457 141) does not make any representation or warranty as to the accuracy or completeness of the statements or information contained in this interview and any liability therefore is expressly disclaimed.
