Even prior to the outbreak of the Coronavirus pandemic the need to automate more of their FX activity was a key topic of discussion amongst many firms in the real money community.
While FX trading desks were often being tasked with handling larger volumes, the size of the desks themselves has not likewise been expanding and so automation was being frequently touted as a means of addressing this imbalance by improving trading efficiency and productivity.
However, as the severity of the current pandemic became clear, causing volatility to explode back into life and market volumes everywhere to surge in response, the idea of automating more FX trading activity has come into even sharper focus. The problem is though, when people talk about ‘automation’ they can be referring to a wide variety of technological tools that can be implemented to varying degrees at different stages of the trade lifecycle. Or, put simply, there’s no one-size-fits-all solution when it comes to automating FX trading activity.
Bespoke execution rules
Most of the time when people are talking about automation, they’re probably referring to systems that automatically execute small ‘nuisance trades’ that land on the desk. This might not sound particularly sexy, but just some very rudimentary math gives an indication of how much time these types of trades can consume.
Assuming it takes on average 45 seconds to complete a normal point-and-click Request for Quote (RFQ) or Request for Stream (RFS) FX trade and an asset manager is executing 100 such trades per day then that’s 75 minutes per day taken up by this activity. Extrapolate that over 250 trading days in a year and it suddenly becomes apparent that traders at that firm are spending a whopping 18,750 minutes, or 312.5 hours, per year mindlessly clicking to execute simple trades — time which could be much better spent on more value-adding tasks. Now to be clear, the tools required to simply automate FX trades below a certain size are nothing new, and in fact are fairly widely available. Instead, what really differentiates automated trading solutions today are the level of customisation they offer and the data set underpinning them.
For example, rather than a basic approach where trades under a certain value are just sent straight out to the market immediately without human intervention, firms might want to configure a set of rules which dictate that a given trade should only auto-execute if there’s a minimum amount of liquidity in the market, or if a certain percentage of their bank basket is quoting within a specific timeframe.
Optionally, they might want to pre-determine that trades should only auto-execute at specific times of the day, something can be especially helpful for firms that have deals coming in from other time zones but don’t necessarily have an overnight desk ready to handle them. Buy side firms might also want to customise the parameters around this auto-execution function to stagger their trades out into the market, effectively enabling them to build their own FX algos. Where this type of functionality gets even more interesting is when it’s applied to NDFs, outright forwards and swaps, given the current paucity of algo options for these products available to most buy side firms today.
The importance of data
It should go without saying that another very important parameter to consider is price sensitivity, and this is the part where having the right data on-hand is absolutely crucial. Firms might want to dictate that trades should only be executed automatically if they’re within a certain threshold of the market midpoint. For spot FX, where there are many options for accessing good quality market data, this is pretty straightforward. By contrast, for non-spot products that don’t trade as much on electronic platforms and where accurate market data isn’t widely available, this is a much trickier prospect.
One advantage that we have at 360T in this regard is our Swaps Data Feed (SDF), which offers full granularity across the curve from O/N out to two years in over 25 pairs, providing data in 400 crosses in G10, EM and NDF currency pairs. This proprietary data set enables us to accurately track the market midpoint for these non-spot products and thereby allow firms to confidently auto-execute based on a pre-defined tolerance range from this mid.
Not only should real money firms be demanding this level of customisation from the automation tools they choose to implement, but they should also be considering whether these configurations can be applied to individual funds within an underlying account. Because while having a set of blanket auto-execution rules across all of their funds is useful, the ability to drill down and create bespoke rulesets for specific funds in order to match their workflow requirements is even more advantageous.
Slowing things down
Even though this high degree of customisation means that automation doesn’t necessarily have to involve giving up control, some firms would still understandably prefer to slow things down at times, and it’s also important for automation tools to be flexible enough to accommodate this.
There are numerous different ways to insert an element of manual intervention within automated solutions. This can involve allowing these tools to be configured so that trades are not executed as soon as they arrive in the EMS but instead rely on a human trader to manually select which trades should subsequently be auto-executed according to the previously defined criteria set by that trader. It could equally involve sending trades to the EMS with identifiers, or handling instructions, that determine whether each particular trade should auto-execute or sit and wait for manual execution.
And even with the ability to dial up or down the level of human intervention required for trading, there will naturally always be compliance considerations when firms look to implement new automated trading tools. And once again, this is where the data under-pinning this automation comes into play.
For firms to be comfortable with increased automation they need to have access to the data which completely validates the trading activity being undertaken.
They also need complete transparency and oversight in the form of audit log reports which cover every single step of the execution process, from the moment when the deal comes in from the OMS to the moment that it matches with a given price.
This data is incredibly important and so it needs to be available to firms using these auto-execution tools through a variety of channels so that they can access these reports in whatever manner is most suitable for their workflow and operations.
More than this, firms need to be able to customise their automated trading tools to ensure that they are entirely consistent with their compliance rules and that these systems immediately raise flags to notify staff immediately if any of these rules are breached. Just as with fully manual trading, there needs to be checks and balances in place so that firms can operate in full confidence that they are adhering to all of the relevant compliance requirements.
Ultimately, the trend in FX — like elsewhere within financial markets — is towards more, not less, trading automation and firms that are able to embrace this change will benefit as a result.
However, the key point here is that the technology and data available today means that buy side firms should absolutely be demanding automation tools that can be adapted to suit their bespoke workflows, are based on the best possible data sources and enable them to remain in complete control of their trading activity at all times.