Building buyside confidence in FX algorithms

June 2023 in Previous Features

Kevin McPartland

“What’s going to incentivise those guys to change their behaviour is some kind of proof that the technology will leave them ultimately with better execution.”

How are algo providers looking to change all that and convince money managers and corporate treasurers to adopt algos as part of their execution repertoire? Different firms are taking different approaches but one thing is clear: the buy-side wants to know more.

“Algos have become significantly more sophisticated over the last 12-18 months, accelerated in part by banks leveraging the expertise of their equities product,” said Richard James, head of Currencies and Emerging Markets Execution Services at JP Morgan.“As a result, client expectations of algos are higher.”

Expected growth

Kevin McPartland, Principal, Market Structure and Technology at Greenwich Associates, said buy-side usage of algos in the FX market is relatively modest but that is expected to change quickly.

“I always found it interesting with FX that, as electronic as the market is, the algo usage is so low. Something like 11% of buy-side firms said they use algos, and we’re talking about a market where 80% of the volume is done electronically,” he says.

“That said, the numbers also showed the biggest expected growth we’d ever seen, something like 7% to 14% by the end of 2014. We’re actually out interviewing now for our next iteration of our annual FX study, so we should find out if that came to be true in the first quarter.”

As could be expected, the hedge fund community has led the way but now other buy-side participants are starting to embrace algos.

“A lot more people are getting to the point where they’re feeling more comfortable in using algos, and the client base has grown from the traditional hedge fund community to include real money and corporates,” James of JP Morgan says.


Greenwich Associates’ McPartland said that for corporate treasurers it may be a bigger hurdle than for other segments of the buy-side.

“There certainly is a push towards using more algos for corporate treasurers and real money. I think the biggest leap is going to be for corporate treasurers. They tend to trade a little less frequently; their needs are more specific, often because they’ve got very particular needs, whether it be hedging needs or business needs. So their trades need to be structured in a very particular way,” says McPartland.

He said corporate treasurers tended to prefer trading over the phone because their trades are less frequent. “It’s much easier to pick up the phone and call if you’re trading once a week or once a month than it is to deal with the platform.”

Real money firms tended to be larger and to trade more frequently. “They’re also used to trading electronically, so using algos is just the next iteration in electronic trading for them,” he said.

Explaining the value of algos and educating clients ultimately can come down to individual users, all of whom have different needs.

“In general, algo use will differ depending on the client type, so it’s not possible to bucket clients into one group and assume that their use of algos would be identical,” James said.

Explaining the value of algos and educating clients ultimately comes down to individual users, all of whom have different needs.

Show me the money

Richard James

“At the moment, the market is still at a half-way house with algos, where clients want to put them to use but at the same time they want to control a number of the parameters. This can, at times, restrict the ability for the algo to work as efficiently as it could.”

McPartland said that from an education point of view, the more providers can show that an algo will produce better outcomes than existing practice, the more the buy-side will get on board.

“What’s going to incentivise those guys to change their behaviour is some kind of proof that the technology will leave them ultimately with better execution.

I think that’s going to be the driver there. I don’t think the automation part really is a driver. I don’t think the process efficiency is really a driver. It’s really all about best execution,” he said.

This was especially the case for real money. “That’s their fiduciary responsibility for their investors: that they’re getting best price and they’re doing everything they can to ensure they’re getting best price. That is what I think the algo should be providing.”

Once a provider makes a convincing case to a money manager about the benefits of algo-based execution, next comes the question of how hands-on the client wants to be in terms of using the algo.

“At the moment, the market is still at a half-way house with algos, where clients want to put them to use but at the same time they want to control a number of the parameters. This can, at times, restrict the ability for the algo to work as efficiently as it could,” James said.

Trading venue complications

Another factor for the buy-side is that different venues have different rules so getting a proper view of liquidity via aggregation may not be as straight-forward as some people think.

“Trading protocols are not consistent across all platforms in the market and so it is important that pre-trade aggregation starts to take into account the nuances of those different protocols,” says James.

What that means is that an aggregator may present an incomplete or misleading picture of the real liquidity situation at any given moment.

Need for more information

Asif Razaq

“We lifted the burden of trying to select the right algo strategy,”

“What clients are very focused on and we are providing is detailed information around how the algorithms work and how they are performing,” says James. The FX space, unlike equities, does not have performance benchmarks to reference algos against, so comparing performance can be tricky as well. Transaction cost analysis, however, is seen as a definite growth area for the FX market.

McPartland says of TCA: “It’s not as big as it is in equities but it’s definitely bigger than it is in fixed income. Certainly there’s a lot of work going on to show, in a quantitative away, that you’re getting best execution or explaining what the total cost of that trade was.”

He added that he did not see safety as a concern for the buy-side. “I think a lot of that’s overblown. I think a lot of the technologies now, especially the algos that are put in front of big real money clients, have been tested for years. And there are error checks upon error checks in there.”

When and which one

Once the buy-side is satisfied that it is in their interest to use algos and once they feel comfortable dealing with the various parameters, there are still nagging questions as to when to use algos versus other methods and which algos they should consider.

“That’s going to continue to be a complex question,” says McPartland. “A lot of it is going to come down to the existing relationships that those buy-side firms have, the brokers that they work with and that they trust the most. Maybe that’s not the most quantitative answer but I think that’s the reality of the market.”

In other words, asset managers are most likely to work with people they already trust and whom they believe will help educate them and act as a partner in their business. “The new providers are not the ones that are going to get the guys who have never done it before. I think the new providers are going to get the guys who already get the joke, if you will, and are already looking for an improvement from their current setup,” says McPartland.

Another group of would-be adopters are those intrepid asset managers who realise that algos are the future.

“That’s what happened in equities as the market automated. It’s not that we lost jobs, it’s just that the job profile changed over time,” says McPartland. “For FX, despite the limited use of algos, electronic trading is very prevalent and widespread. So, this is a group of traders who understand trading electronically and RFQs and clicking on bids and offers, so they understand how to interact with the system and this is just another tool, or another layer added on top to help them get their job done,” he said. “It’s not a quantum leap.”

What people want is factual information about how algorithms work

Demonstrating the value

BNP Paribas is one example of a firm that has wrestled with the question of how to demonstrate the value of algos while making them understandable. There, the emphasis is on simplicity.

“One of the key principles for iX, our algorithmic execution service was to simplify the product. We achieved this by keeping the family of algos very small and naming them accordingly so it’s quite clear what the algo is designed to do,” says Asif Razaq, global head of FX algo execution at BNP Paribas.

The algo designs are based on the degree of risk the client wants to take. If the client is prepared to take on some risk, then it could opt for an algo such as BNP Paribas’ ‘Chameleon’ algo, which is designed to focus on adapting its strategy to market conditions. Or, if the customer wants to minimise risk, the bank’s ‘Viper’ algo will execute more quickly.

“We lifted the burden of trying to select the right algo strategy,” he said, noting the bank’s Chameleon algorithm will make adjustments based on real-time analytics.

“With conventional second generation algos, clients would have to assess market conditions and determine the right algo strategy,” says Razaq. And then, if market conditions changed, clients would need to make a switch and take a different approach.

In terms of education, Razaq said BNP Paribas has conducted roundtable discussions where it has invited in clients. He says different clients will have different sets of questions.

But the biggest question mark BNP Paribas sees is when to use algos in the first place. To address this, it plans to roll out a new pre-trade analytics service (see news section) to help customers make an informed choice.