Andrew it’s been 5 years since we last did an interview you so please remind our readers about your day-to-day responsibilities?
As Vanguard’s Global Head of FX, I am responsible for all of the currency needs for the franchise. This includes executing all of currencies orders for our mutual funds and ETFs, over 3 trillion dollars per year. I also lead Vanguard’s currency alpha generating strategies for our active global fixed income portfolios. I manage FX traders in the United States, United Kingdom, and Australia. I also serve as a senior advisor for Vanguard’s Women Inspiring Leadership Success and as a Co-lead of Vanguard’s Black Professional Network’s onboarding and attracting arm.
How is the global FX trading operations at Vanguard structured?
We are a global desk with FX traders located in 3 offices, United States, United Kingdom, and Australia. Even though traders are located in different regions we are a centralized desk that is agnostic of region. This structure provides 24-hour 6-day-a-week coverage of the FX markets. With dedicated FX traders in these regions, we can pass orders around the globe and take advantage of liquidity pockets that exist when local markets are open. We can easily match order objectives by handing off orders to other regions based on need.
Vanguard manages assets all over the globe and you hedge many products to multiple currencies. How do you bring all of those flows together to get a sense of what your global exposure is?
We use a combination of proprietary as well as external order and execution management systems to aggregate all of Vanguard’s currency needs. We are extremely focused on best execution, minimizing impact and reducing transaction costs. Via our systems, we identify every opportunity to net and aggregate FX orders. Since the inception of Vanguard’s FX desk in 2012, we’ve netted over 1 trillion dollars, saving our clients hundreds of millions of dollars in the process.
Broadly speaking what are the main objectives of your FX trading activities and what factors influence how you go about achieving them?
The vast majority of our currency needs come from passive multinational portfolios. A US domiciled Japanese equity fund receives US dollars from a US investor. That fund then buys Japanese equites, a currency transaction is required to turn US dollars into Japanese Yen so the fund can settle the Japanese stock purchase. We also are responsible for hedging our fixed income portfolios from currency volatility.
The main point of these two examples is that we do not have a view into the currency. We are trading to fund an equity purchase on trading to immunize a fund against unwanted currency volatility, not because we believe Japanese Yen is going to appreciate or depreciate in value. The passive nature of our flow allows us to focus on the overall cost of trading. We do not have any urgency or potential alpha decay, therefore we can take our time trading and make sure we are receiving the best price possible while minimizing negative impact.
Vanguard is one of the world’s largest investment companies with around $5.6 trillion in global assets under management, as of August 31, 2019
Do you think the job of FX trading has become more complex than it was just a few years ago?
FX market structure continues to evolve and mature and the number of tools a trader needs to ensure best execution continues to increase as well. Liquidity is fragmented across 100 different venues and liquidity providers, new regulations have constrained banks’ balance sheets, risk appetite to hold FX market volatility has diminished.
These factors helped create the need by real money shops to focus more attention on FX execution. If done well the FX market can be a real source of competitive advantage as the ability to save on execution costs continue to increase.
In what key ways has increased automation impacted on the way you and your teams work and what benefits has it brought to your trading and workflow processes?
We are only at the beginning of the positive impacts of automation but already we’ve experienced huge efficiency gains. Automation of small orders and routine tasks free our traders time so they can focus on larger orders that could potentially impact the market. Execution savings on these orders are meaningful.
Let’s talk a little about FX algos. How much of trading now involves the use of FX algos?
FX algorithms are an extremely important execution tool for us. We were one of the first real money shops to start using FX algorithms. We’ve partnered with our banks from day one to create dynamic execution algorithms that do a great job staying anonymous, accessing liquidity, and reducing impact. Over 99% of our large spot orders are executed via an algorithm. The savings has been significant.
What type of FX algorithm are you typically employing and what issues generally influence this?
The specific type of FX algorithm a trader decides to use depends on important factors including the benchmark, order size, currency pair, liquidity, and urgency of the FX order.
Today there are algorithms designed to handle almost every need a trader has. We tend to be extremely passive with our algorithmic orders, opting for less impact and more spread savings.
We are a global desk with FX traders located in 3 offices, United States, United Kingdom, and Australia
Your background was originally in equities. How helpful was your experience in using algorithms in that market when you moved across to FX?
I think that experience was everything and one of the main reasons our desk has been so successful. When I launched the desk, I knew just enough about FX voice trading to be dangerous. I did however know a lot about execution and algorithms. At the time, FX market structure was starting to look more like equity market structure than fixed income market structure.
The team’s equity experience allowed us to hit the ground running. We worked hard with our liquidity providers, execution management platforms, regulators, and transaction cost analysis providers to create a fair, transparent, and accessible market place for all participants. We drew on our equity experience to avoid mistakes we had witnessed in the equity market structure evolution. We made recommendations that added value, which included features to FX algorithms. Simple things like parent and child orders didn’t even exist when we first started trading. I am extremely proud of the contributions our team and our partners made along the way. We contributed to making the FX market a better place for everyone.
How do you source your FX algos and what factors are important in deciding the best providers to deal with?
We deal exclusively though bank-sponsored algorithms, meaning we don’t create our own. This means we don’t need to have a direct relationship with all of the different ECNs and we can face our banks directly via our bi-lateral agreements.
We also don’t need the significant resources that writing FX algorithms requires and we can rely on our banks to vet each ECN’s liquidity profile. Without significant data points and observations, it would be difficult for us to accurately map the deepest and cheapest liquidity at any given time. Agency style algorithms align our and our banks best interests; these algorithms work well with our execution style. We’ve worked with many banks to provide us with beneficial features such as bespoke liquidity, peer matching, and pure passive strategies.
FX algorithms are an extremely important execution tool for us
How involved do you get in the testing and fine-tuning of algos during the development process and in what ways do you try and benchmark their performance to see what value they are having?
We are heavily involved in the development process making many suggestions and recommendations ranging from routing logic to peer matching. Once an algorithm is ready for production, our traders will begin using it; from there we use TCA systems to help identify the quality of the algorithm. We determine when and how each algorithm should be used. TCA reporting helps to identify algorithms that work best under different circumstances.
What sort of metrics are you looking for from your TCA reports.
We have an entire team that is focused on TCA across asset classes. This is extremely beneficial because TCA has evolved so much over the last few years. TCA is a full-time job. We look for many of the same metrics that we do for other asset classes, sucah as, slippage from arrival-mid and estimated risk-transfer-price. We look at each fill over the life cycle of the trade and the fill price against market-mid at the time of execution. Market impact is also very important, we measure if our fills were aggressing through the bid-offer spread and if there was any reversion after our order was completed.
What types of FX orders are usually are usually a good fit for algo execution for your trades?
Large passive orders that do not have urgency tend to be best suited for algorithms. We would not want to pay the extra premium to a bank for a risk-transfer price on an order that does not have alpha. Instead we will take on the market risk and use a passive algorithm that will slice the large order into many small pieces. Over a long period of time the market gain or loss nets to zero and we always save on the execution costs. These savings are significant, not only do we not pay the premium of the risk transfer price, we also save against the arrival-mid-price. We do this because we stay on the passive side of the bid-offer, essentially being paid to provide liquidity to the market.
Some firms prefer to adopt a hybrid human approach when using FX algorithms which involves laying down tight limits for their use. How willing are you to let algos do their job without micromanaging many aspects of the execution process?
It really depends on the why we are using the algorithm in the first place and what our objectives are. We have dedicated FX traders that monitor everything while an algorithm is running. We tend to intervene a lot depending on market conditions.
Randomness is a good feature to have in an algorithm and nothing is more random than human behavior. Using limits is a best practice for a number of reasons. We always use limits but at the same time we want the algorithm to work as it was designed.
Do you have any concerns about the potential risk of FX algo trading? For example with algos that don’t have proper constraints and controls or conflicts of interests that arise with providers.
We view FX algorithms as another tool available for execution. The FX trader is the one responsible for best execution. Limits should always be used and traders need to monitor the algorithm while an order is being worked.
Running a large order via a TWAP algorithm during an illiquid time can be disastrous but that would be the trader’s fault and not the algorithms. Banks have a wall in place to ensure algorithm information doesn’t get shared with their principal desk. The banks we deal with do a great job, if they didn’t, we would start to see slippage to arrival-mid while using their algorithms.
“I am extremely proud of the contributions our team and our partners have made towards making the FX market a better place for everyone”
What would you like to see done to improve current FX algo and associated TCA offerings?
Data and integration with TCA venues is the area that needs the most improvement. We are seeing vertical integration between EMS and TCA providers. That has alleviated some of the data transfer issues but more work is needed. Standardization of reject codes across the industry would help identify the costs of a rejected order. This would help the industry understand the true cost of last look.
Ultimately what are the main benefits big firms like Vanguard get from using FX algorithms?
Higher returns for our shareholders. We’ve added over a billion USD of return to our portfolios by saving our shareholders on transaction costs.
The FX Global Code and regulatory developments are driving increased demand for FX algos and TCA amongst many buyside firms. What impact have they had on your own operations?
We started using FX algorithms well before the FX Global Code. Algorithms always made sense for our FX orders.
Algorithms make even more sense now because banks no longer warehouse risk for very long. The benefit of a risk transfer price doesn’t exist. You can save costs by warehousing the risk yourself and executing through an algo.
Is Vanguard likely to expand its use of FX algos in the future.
Yes, we are looking to add new products like non-deliverable-forwards and potentially FX swaps to the algo trading mix. We believe we could see similar savings in these traditionally high-touch-voice dominated instruments.
We are also seeing more peer-to-peer options. FX Hedgepool allows us to match our swap flow against peers going the opposite direction. This could be a new source of liquidity that will complement bank liquidity.
FX algo trading is becoming a technology arms race. Do you worry about the rise of the machines learning and AI technologies for your long term job prospects or will FX always need professional and highly experienced human traders?
We are long way away from human-less trading desks and we are excited for what technology will bring. Automation will allow our FX traders to focus on larger value add orders. Ultimately humans will always need to be involved in the process, technology will make trading jobs easier by creating efficiency.