Adapt and thrive: how FX algos are coping with volatility

Vaovao ara-bola

COPYING AND DISTRIBUTING ARE PROHIBITED WITHOUT PERMISSION OF THE PUBLISHER: SContreras@Euromoney.com

by:

Paul Golden

Navoaka tamin'ny:

FX trading algorithms are getting smarter at dealing with crises – and getting more popular as a result.

A few weeks is a long time in a volatile foreign exchange market. 

At the end of March, Greenwich Associates published a report entitled ‘Digitization delayed: why algos aren’t more popular in FX’, based on data collected before markets felt the impact of Covid-19. 

It found that only 37% of FX market participants surveyed used algorithms – and just 22% of their volume was traded algorithmically.

Fast forward to late April, and JPMorgan’s FX e-commerce team was telling clients of a big increase in volumes of algos for tickets with a notional value above $10 million, with 60% of these orders traded algorithmically in March.

Jill Sigelbaum,
FXall 

Buy-side traders working largely from home while coping with a significant increase in the number of trades to execute have turned to algos so they can focus on more time-consuming, less liquid and complex trades.

Jill Sigelbaum, head of FXall, notes that algo trading volumes on that platform increased by 380% in March compared to the same period last year, with the bulk of the increase attributable to asset management clients.

“We were told by some of our clients that with the challenges in trading from home with less desktop real estate, they were appreciating one-screen [meaning one-system] access to multiple algos across multiple banks, as clients had to restrict the number of applications they had open at any one time due to corporate VPN restrictions on bandwidth,” she says.

Increased familiarity

According to Ralf Donner, head of FICC execution solutions at Goldman Sachs, client demand for execution in larger sizes than usual is another factor behind recent growth in algo usage.

Ralf Donner,
Goldman Sachs 

“Large algos had been tried and tested in calmer markets and were given the benefit of the doubt in the more volatile period,” he says.

Increased familiarity with the technology is also a consideration. Compared to previous periods of high volatility (such as the EU referendum and the 2017 US presidential election, for example) clients now have a better understanding of how algos work.

In addition, the volatility in March was over a sustained period rather than a singular shock or focused on a specific date, explains Scott Wacker, head of FICC e-commerce sales and marketing at JPMorgan. “Clients therefore had time to observe and change their approach,” he says.

The perception that algos don’t perform well in volatile markets is largely derived from the experience of flash scenarios, where spot moves dramatically and algos trigger their circuit breakers. However, Pete Eggleston, CEO of BestX, says his firm’s research indicates that not all algo styles are unsuited to more volatile markets.

Scott Wacker,
JPMorgan

“Using arrival price as a benchmark, our analysis shows that ‘get done’ algo styles [the most aggressive algorithmic order type] improved their performance in the first quarter of this year compared to Q4 2019,” he says.

FX markets have been mostly well-behaved during the crisis so far, despite wider spreads and greater volatility. The main challenge for algo execution, therefore, has been to re-route orders automatically to the available sources of liquidity – among which internal matching was very important – and to minimize the parameters governing execution.

“Perhaps it wasn’t the ‘fire and forget’ approach that algos had become for many clients pre-crisis, but execution outcomes compared to typical benchmarks were still roughly normally distributed – albeit with a wider standard deviation,” adds Donner.

Adaptive algos

That might be partly down to greater use of adaptive algos. Asif Razaq, global head of FX algorithmic execution at BNP Paribas, says adaptive algos can adjust their execution behaviour to fit the market, becoming more or less active and chasing pockets of liquidity.

Asif Razaq,
BNP Paribas

He says that clients not only save on the spread quoted by their banks on a risk transfer basis, they are also able to capture that spread through an algo by posting interest into the market.

“When spreads are several times their normal values this can add up to a considerable cost saving,” he says.

A review of emerging themes and challenges in algorithmic trading published in April by the FICC Markets Standards Board observed that use of execution algorithms in FX was underpinned by an increasingly fragmented market structure where liquidity can disappear quickly.

Some think the trend for greater use of algorithms will outlive the coronavirus volatility, as traders will continue to value the attractions of lower spread.

“If they have reached the conclusion that that is a good thing to do in volatile conditions, they are likely to reach the same conclusion in quieter periods,” says MahiFX CEO David Cooney.