There is no such thing as zero touch trading

You might be cruising on autopilot for most of the flight, but when there’s turbulence you want a pilot to take over, writes Annabel Smith.

The growth of electronic trading capabilities is so ingrained in the evolution of the markets, it’s almost not worth pointing out. We live in an era of rampant innovation and as technology has become more advanced, traders have increasingly lent on it as a means to improving their trading experience and, most importantly, their results. 

As best execution has become front and centre, electronic trading offers a means of quantifiably measuring and thereby holding accountable the institutions guarding the purse strings. And that is a good thing. The developed listed equities space has paved the way for other asset classes in this regard, with an increasingly large portion of volumes in this market now accounted for by trading engine to trading engine activity. Smart order routers sit downstream of algorithms that interact with one another on an order book in real time.

Artificial intelligence is playing an increasingly important role in almost all aspects of life – even journalists are cautiously eyeing ChatGPT, feeling a prickle of sweat every time someone shares another flawlessly written article – but when it comes to buy- or sell-side traders taking their hands off the wheel completely and allowing orders to auto-execute with no supervision through zero or no touch trading, the financial industry is not yet ready or willing. Not least due to the volatile market conditions witnessed in the last few years thanks to unpredictable events such as the pandemic and the Russian invasion of Ukraine. You might be fine cruising in auto pilot for most of the flight, but when your plane hits turbulence, you want a pilot to take over.

It boils down to risk 

For much of the traditional long-only buy-side in particular, zero touch trading increases risk for very little return. Unlike proprietary trading firms, asset managers are accountable to their portfolio managers and ultimately their end investors. In today’s increasingly transparent and TCA driven environment letting technology have full control without any oversight, is a fool’s game. Instead, most favour a one or two touch approach. One for an extremely liquid stock or perhaps if an order is generated off the back of another.

“Start with two and when you get confident you go to one. I wouldn’t sleep well at night in my job going to zero,” says head of trading at Bluebay Asset Management, Stuart Campbell. “If you had tight enough controls you could have zero touch but you’re opening yourself up to a lot of risk you don’t need to take.”

The reality is, a completely zero touch system with no checks or monitoring in place is ladened with risk that buy-side firms just can’t afford to take. It’s reliant on airtight data – something still on the wish list of the buy-side, particularly in fixed income – and processes in place to deal with any outliers or unexpected activity in the market. What’s more, the larger a firm is on the spectrum of assets under management across its business, the more careful it has to be with ensuring it’s got good internal policies in place around aggregating and crossing orders to ensure there’s no conflicts. Technology is a tool to be used by traders to reduce clicks but not to withdraw all together.

“There’s no such thing as no touch right now and I don’t think there ever will be,” says senior analyst, risk and financial markets regulation at Coalition Greenwich, Audrey Blater. “Let’s say you’re an investor and want to put your money in a fund, I don’t think it sounds like a good risk practice to say we have zero touch trading.”

Ongoing squeezes on supply chains thanks to Covid and the Ukraine crisis have left most countries around the world on the precipice of recession. Cost reduction now more than ever is at the forefront of the minds of those overseeing budgets bringing to the fore new discussions around outsourcing either via outsourced desks or, technology. Some institutions are happy to pay as little as 0.1 basis points for an extremely low touch service, however, in exchange for this low fee how much risk are they agreeing to take on?

Value add

If the market were to reach a point where trading involved no touch or eyes from either the buy- or sell-side firm, it also brings into question what investors are paying for when they part ways with their cash in exchange for their services. With the prospect of outsourced trading looming, all parties must prove their value add to the trading process. Looking at it in the most morbid sense, a key driver for high client fees is that a human service offers firms someone to hold accountable when something goes wrong. This knowledge allows firms to entrust huge amounts of capital.

“One touch gives clients security that you have a set of eyes on it. It’s imperative that it’s still a decision made by you and orders are grouped accordingly by you,” said one buy-side head of desk.  “There is always an element of value to be added, I can’t see how people can be fully automated because of the diverse nature of clients’ needs; mandates, funds and individual rules of the IMA and prospectuses. There are too many nuances to completely automate away this risk. The fund management industry has multiple parties to respect and important regulations which must be adhered to.”

On the sell-side, amid flat volumes in Europe, particularly in equities where competition is high and commissions are low, market conditions have meant client offerings must now offer value add around a growing list of factors to prove their worth over their peers: including customisation to specific limits or types of liquidity, managing market impact and handling around connectivity or market outage issues. Despite an order being automated to the point that no trader has touched it, most firms will agree there is always some level of involvement. Monitoring orders for outliers is essential and there is always a kill switch. 

“We have people now that are completely dedicated to algo wheel monitoring and performance enhancement,” says head of EMEA electronic and program trading at Jefferies, Ben Springett. “Often it’s the outliers that you observe in real time that give you information about how you maybe adapt that strategy to avoid those occurrences repeating in future, and you need to be fast. Often you have just a quarter of measurement that determines whether you stay on the list, do you get smaller allocation or bigger allocation?”

As the markets become increasingly complex around regulation, reporting and fragmentation and as algorithms too have become more advanced, the role of the sales trader and trader has advanced as well. Historically a client might send an order into a liquidity seeking algorithm in a process that to the naked eye might seem simple, but Europe for example is now pushing almost 30 venues. What’s more, as strategies have become more complex, firms feel more confident in sending larger orders into automated strategies and this has increased the need for a hands-on approach from sell-side counterparties.

“Coverage is involved during the life of a trade and post-trade. It is about how those algos performed, how can we do better, what we should be looking to build and how we can continue to innovate. This side of the business has become ever more complex, we need people who are both technical and personable,” says Bianca Gould, head of electronic execution at RBC. “There is an expectation that we are overseeing orders. I don’t think zero touch can truly exist. Clients expect us to add value. We have an AI product where the algorithm takes ownership of that order effectively but that doesn’t negate the need for interaction and for coverage to be very much on hand.”

Naked Direct Market Access (DMA) is available to the buy-side, sidestepping the role of the sales trader and instead allowing the firm to utilise a broker’s pipes and infrastructure to directly access an exchange. However, most brokers would argue that this model alone often leaves buy-side firms vulnerable to fat finger – market impact – and that DMA should be overlaid with algo suites to avoid being subject to market moves against an order that hasn’t been properly placed in the market.

There’s always a touch somewhere 

The role of the trader and sales trader has evolved significantly. Whichever way you look at it they have become more hands on as opposed to less, in particular in the post-trade process thanks to the TCA-driven environment the market has morphed into. The intensive process of monitoring performance has replaced the formerly manual process of trading. 

Trading is without a doubt far less touch nowadays in terms the actual physical process. The pandemic was the final nail in the coffin for voice trading and for the most liquid stocks it is a thing of the past. Up to a third of Jefferies’ volumes are accounted for by so-called “black box” trading whereby a firm will have no individual trade inputs by a human being on either idea creation or on the subsequent order that’s generated or its execution. To another degree firms can code investor strategies that are then given a license to execute against that investment thesis on a fully automated basis. 

Further levels of automation vary across algo wheels whereby an order goes into a wheel that selects the appropriate pre-defined broker algorithm and executes it without touching the desk. No human has touched that order since making the decision to generate it and send it there. 

“If a broker is algo pricing my enquiry then that could be zero touch. For me to generate a ticket on my side I use controls, checks, balances, integrated mandate rules etc. It’s one touch for the buy-side and zero touch algorithmic on the sell-side,” adds Campbell. “I want to check a trade before it executes automatically. Most buy-side shops are doing the one or two stop checking process. On the sell-side if someone enquires about buying a stock and they have it on their book and their algo sees it and knows it can do the best price and gives it, they don’t touch that.”

Systematic and program trading firms that use the speed of their proprietary algorithms to gain efficiencies in the market are the most advanced when it comes to automation. But again, it must be stressed, someone, a human someone, has to understand the trade, understand its counterparty limits and what’s happening in the wider market. 

“That’s [program trading] a speed game, machines are faster than people but they only make the decisions that people programme into them. Somebody has to be watching and they are,” explains Blater.

In the listed equities space, the most outsize returns have accrued to those people that were first on board with digitisation and automation. A growing portion of what the largest institutions execute is done via black box methods of trading. Consolidation and economies of scale have meant the largest firms have continued to get larger. 

These are the firms that benefit most from zero touch trading, explains MarketAxess’ global head of trading automation, Gareth Coltman.

According to him, demand from the largest institutions for zero touch trading services has continued to grow in the last 12 months, as a means of easing the pressure off of desks that have a growing workload and a shrinking headcount.

“That isn’t to say there isn’t a trader supervising that process, finding that workflow, helping to define those rule sets and dealing with exceptions and orders that don’t fit into that workflow,” he says. “Having the ability to use a tool like this when the desk is very busy is very appealing. For some firms it’ll be something they use every day or for others when they’re particularly pressed for time. Every desk has to assess how much benefit and added efficiency they get from an automation solution and that benefit is likely to correlate with volume of activity.”

MarketAxess recently launched its Adaptive Auto-X service which aims to make plugging into fragmented liquidity pools easier for users by taking MarketAxess’ liquidity pools and placing them alongside its data and analytics offering, allowing them to set up more complex automated workflows.

Various other firms have come to market as of late with products advertised as zero touch including fixed income artificial intelligence quantitative analytics provider, Overbond, which announced a new integrated AI-driven margin optimisation model in October that should allow for end-to-end zero touch automated trading. But, exciting as it might sound, these systems are not acting alone.

“Sophisticated investment firms react to exceptions and enhance strategies and have clever quants to constantly make calibration changes to the way the machine is modelled,” says Andrew Morgan, president and chief revenue officer at TS Imagine. “To that end, these businesses are hands on and high touch.”

The argument can be made that the further you go downstream in a trade the more no touch it becomes. VWAP strategies executing an order over a period of time will slice it into smaller micro-orders that AI micro traders will then distribute to various liquidity pools or strategies. However, step back to examine the larger order and you’ll find a person has parameterised that VWAP algorithm. As trading stands today, there is a person (if not three) involved somewhere in the lifecycle of a trade, including its execution. Automation is at its most powerful when it is augmented with the human trader. 

“It’s nice to say “no touch” but I think it’s a misnomer. It’s not that no one’s touching it they’re just touching it less,” says Blater. “I don’t think zero touch is a realistic goal. I think some firms use interesting marketing language around automated features and rules-based trading but really that’s just a technology to enhance what the human being is doing. I don’t think the goal is to reach no touch.”

«