One of the main challenges that comes with FX algo execution is that clients may have to jump in and suddenly be spot traders, despite not having that degree of familiarity with the market, says Donner. They certainly, in general, would not know what live conditions are like in the spot market or how to translate current conditions into an informed execution decision and very often they will have a choice of execution options, he adds. “They might need to decide whether to opt for an opportunistic or passive algo, or use risk transfer or even sweep the book to fill instantly,” Donner says. “The question is, how to choose? There are a wide variety of pre-trade tools which are all attempting to help clients address this question. What we discovered, however, is that in general these dashboards of tools, only address one specific aspect of the problem, but nothing is able to address the question in its entirety.”
Donner explains that a traditional pre-trade tool might show individual factors such as market risk, volatility, what live volumes are like versus normal, what the spread is versus usual conditions, whether to execute faster or slower etc. He adds: “This means that clients have lots of little dials and gauges that they need to look at to make their decision. What was not available yet was something which can bring all that data together in a single modern display that immediately and visually addresses this question, so the client can just look and confirm that yes, under these circumstances, it should be an algo, or it should be risk transfer and if it is algo, then this is likely outcome, and if it is a risk transfer, then this is the outcome – to be able to simply compare those at a glance.”
Having recognised there was a need for this level of next gen pre-trade tool, Goldman Sachs has developed its Visual Structuring Delta One (VSD1) tool which it is about to make available to clients. Churchman explains that the inspiration for this new visual structuring tool stemmed from his experience in creating a structuring tool for mainly vanilla FX options structures. “It turns out that there are such incredible similarities between structuring an options trade and choosing a spot execution strategy. It then seemed a natural progression to extend this tool for use from the options world to the Delta One world,” he says. “They are both structuring problems with payoff profiles, a dependence on market volatility, and parameters that influence possible payoffs, such as strikes or a limit price.”
THE INSPIRATION FOR THIS NEW VISUAL STRUCTURING TOOL STEMMED FROM THE CREATION OF A TOOL MAINLY FOR VANILLA FX OPTIONS. THERE ARE SIGNIFICANT SIMILARITIES BETWEEN STRUCTURING AN OPTIONS TRADE AND CHOOSING A SPOT EXECUTION STRATEGY.
Live market conditions
Churchman adds that the tool essentially displays the current state of the FX market, as represented by an aggregate view of the order book. He explains: “The graphic then shows what is out there in terms of how far through market the client would need to go in order to fill various sizes, specific to the order that the client intends to execute. The tool then displays what is going on in the market right now, where the client would have to execute the order if they wanted to execute quickly.”
Algos, however, have more of an uncertainty around the execution outcome, Churchman says. “We needed to represent that uncertainty using a volatility cone that goes into the future, showing the expected market risk that the client is taking on an algo execution – expected because of market moves and the variable duration of opportunistic algos. We are then comparing a band of possible results with the certainty of a risk or an estimated sweep execution. The more the distribution of algo outcomes lies beyond risk transfer, the more this favours risk transfer, and vice versa.”
This volatility cone is generated using a realised form of volatility which essentially looks at recent spot returns to produce the cone. Live conditions have a direct impact on the volatility projection. Donner notes that as the entire tool is live this can change depending on current market conditions. “This is a very fundamental difference with existing pre-trade tools, many of which have approximately the same value as receiving yesterday’s weather forecast when planning a journey today. What you actually want is something that would give you live information now to be able to plan whatever it is that you’re doing. And that is essentially what this VSD1 tool achieves. It is not easy to pipe live conditions on market liquidity to clients, and so very few tools do so effectively.”
Stream-lined decision making
The reason that this resonates with clients is because of the growing requirement to demonstrate best execution to their stakeholders, which can be harder to achieve when executing with algos, explains Churchman. “This tool now allows them to easily show the justification for using an algo to their client, or to portfolio managers or even for compliance purposes,” he adds. “It can also be used to gain experience with a product and be able to plan for the future. It takes time to build an understanding of how various algos perform, such as the typical algo duration for a certain currency pair. If a PM were to ask an execution desk how long it would take to execute their orders, this would have previously required the desk to remember how long that size in that currency pair at that time of day usually takes with a given algo, and then whether current conditions might hasten or slow execution relative to normal. VSd1 provides the information directly.”
According to Donner, the new tool will initially be available for the most popular algo strategies offered by Goldman Sachs, namely its flagship Dynamic Hybrid algo and the Pegged algo. It can however be used with other strategies, he notes, such as Goldman’s TWAPs which are more customisable than many other TWAPs on the market. “In particular, we allow clients to specify how each clip of the TWAPs is executed, whether that clip should be very passive in nature, or whether it should be hybrid in nature,” he explains. “Because a TWAP can be chosen to be Pegged or Dynamic, a logical way to determine a TWAP end time would be to test the estimated duration for the same size with a standalone Pegged or Dynamic algo…So it does have additional application to scheduled algo types as well”.
