An unexpected love triangle!

June 2023 in Previous Features

After a period of extensive media attention on the traditionally opaque and free-wheeling foreign exchange industry, FXAlgoNews explores how the regulatory landscape could affect the trend of algo adoption by the buy-side.

Conventional wisdom holds that markets have a distinct distaste for regulation. But for the FX industry, is it possible that moves towards more market regulation and closer scrutiny of FX trading could end up helping increase adoption of algorithms?

That seems to be a view among some industry watchers, who say that the broad push in the regulatory community towards fostering best execution may ultimately encourage more take-up of execution algos. This is even the case despite all the attention on FX fixings, which, it turns out, could ultimately also promote algo usage. Industry executives stress that regulation is not a direct driver of algo adoption, but it could be having an indirect effect.

Jim Cochrane, a director and senior product manager for post- and pre-trade FX TCA at ITG, said that when it comes to execution, it’s difficult to see how algos wouldn’t generally outperform humans.

“I’m a strong believer in the future of algos,” Cochrane said. “A human being will have a difficult time outperforming an algo. In my mind there’s very little doubt about it.”

Cochrane, who has a background flying spy plane missions during the Cold War, is someone who has been trained in the value of military precision. It is with that context that he noted how TCA has the ability now to measure on a millisecond basis how well an algo performed against expected spreads – and hence expected cost – of a given trade. “That’s what I am actually showing my customers now. We’re able to take a look at the tradable quotes in the market and come up with foreign exchange cost curves.”

Those curves, he said, will show what you could have expected to pay if you traded on an ECN or if you traded voice in terms of how many basis points away from the mid-price of that expected spread. “The algo traders in the end will be able to show that they are doing a better job versus those benchmarks,” he said.

“I’m a strong believer in the future of algos,”

Plenty of people in the industry have made the case that algos offer value; but how is it then that regulation and closer scrutiny will help people see that value?

“One of the areas of MiFID II that everyone talks about is best execution. That’s definitely something that’s on the table,” said Sang Lee, a cofounder and managing partner at Aite Group.

Lee notes that the FX market doesn’t have benchmarks, which leads to the question: what does best execution mean? “FX algos are being pushed as tools to at least facilitate best execution,” he said.

At the same time, the algos help investors perform transaction cost analysis. “As you leverage an algo you can capture a certain amount of an audit trail,” Lee said.


The irony is that MiFID II does not even directly address spot FX.

Since the global financial crisis, regulators have been a busy group. The G20 Pittsburgh summit of 2009 kicked off a campaign of measures large and small to tackle all manner of issues, particularly in the OTC markets.

Yet, foreign exchange trading, the biggest and deepest OTC market of them all, has been relatively less affected. For instance, while various FX derivatives are included in the final technical guidance for MiFID II, officials have said spot FX was not in scope.

And spot FX represents a very large share of the global market, at some $2 trillion a day in the latest BIS survey and not far behind the leader, FX swaps.

Cochrane’s colleague Michael Sparkes, director of analytics at ITG and the author of a recent paper on multi-asset best execution in light of MiFID II, says that in one sense it

is surprising that spot FX in particular, being such a large and important market, wasn’t included in the compass of the regulation.

“But I think certainly buy-side market participants are treating it as if it was,” he said.

“FX algos are being pushed as tools to at least facilitate best execution,”

In other words, it’s the overall direction of the regulatory focus that is having an impact and encouraging investors to consider algos.

“Buy side clients have already started to see the benefits of systematic measurement and monitoring of their implicit trading costs in FX markets, and where they’re outsourcing it, to have good accurate measurement of a third-party’s performance,” Sparkes said. “So I think that in a sense the market in this area is jumping ahead of the regulation.”

He described MiFID II in that regard as a sort of parallel set of developments. “You can see the direction of travel, shall we say, in terms of regulation but also in terms of market practice. In all major asset classes, the investment industry is looking for ways to have better data, better transparency, but at the same time not compromising liquidity. I think that’s the challenge in the OTC markets.”

Just as the first Markets in Financial Instruments Directive (MiFID) had the unexpected consequence of ushering in the rise of equity algo trading, the second version then could help promote automated FX trading.

There is of course a difference between the two landmark regulatory efforts. In the case of the first, it came about by fragmenting liquidity and creating greater opportunities for algo-based arbitrage. In the case of MiFID II, it comes from a determined focus on the concept of best execution, something execution algos are designed to achieve.

Justyn Trenner, Global Head of Strategy at EBS BrokerTec, said that where regulators have shown concern about algorithms tends to be when they are used for high frequency trading, whereas the kinds used by asset managers are simply about obtaining the best possible fill for asset owners. “I think there’s a big difference between those kinds of algorithms and the kind of algorithms that are used in some high frequency trading techniques.”

Trenner said one area where regulators did seem to be taking aim, particularly in Germany but with the potential for more traction in Europe, was the idea that algorithms needed to be “babysat”, in other words that firms didn’t rely solely on a so-called black box to do execution but rather that a human was always at the switch and responsible for any automated trading.

“I think that that closer link between people and their algorithms is of huge benefit for the market. It means the sensible and safe use of algorithms,” he said.


While spot FX may not have been included in MiFID II – or in Dodd Frank for that matter –

there is little question that there is increased focus on it following scandals surrounding the WM/Reuters fixing.

Even here, however, adoption of algorithms could benefit in the long run.

Simon Wilson-Taylor, Trenner’s colleague and the head of the newly created EBS Institutional (see news item in this edition), said he had never heard from a client who said regulation was a driver of algo adoption. But he added that as a consequence of regulatory findings around the WM/Reuters issues, some banks have retreated somewhat from being as willing to do large risk transfer trades.

“The customers had to find other ways of executing those trades, and one way is to actually manage the risk yourself rather than just doing a risk transfer,” Wilson-Taylor said.

Once a firm begins to manage the risk itself, it needs to figure out how to build a position. At that point, an asset manager is likely to think about a systematic method for building a position because that’s something that can be shown to clients as being consistent and repeatable. “So I will use an algo. That’s something I can point to and show I have a process,” Wilson-Taylor said. “I think it’s a consequence of what’s happening at the banks.”

The fixing scandals have made some firms more wary in their approach to FX trading, although there is a view that it has only slowed what algo advocates say is an inevitable trend.

“Prior to the scandals we assumed a very rapid growth rate in algorithmic trading because it just makes sense. It’s a tool that you would want to use to improve your pricing as you manage that currency risk,” says Cochrane.


“Buy side clients have already started to see the benefits of systematic measurement and monitoring of their implicit trading costs in FX markets…”

Cochrane believes the introduction of the Volcker Rule, which is meant to promote an agency-based rather than principle-based role for banks, is bound to have an effect on the FX market. If banks can no longer hold risk on their books, that has to encourage an agency model.

Sparkes said that after some of the experiences of the recent past, there will undoubtedly be an evolution in the market structure.

“And with the banks under pressure as far as the ways that they can use their balance sheets are concerned, I think the market is still evolving and looking for ways to adjust and to mitigate the issues. Whether it’s in currency markets, whether it’s in fixed income markets, I think that you can see a period of quite rapid evolution over the next three or four years,” he states.

“That has implications for buy side and sell side. I think it will lead to quite a different market structure in the latter part of the decade, where things like algos or crossing networks or dark pools will certainly play a part. But I think that equally the potential for the buy-side community to collaborate in new ways is also potentially part of that new picture.”

He said that in that sort of environment, one could expect to see new ideas being driven by either the buy- or sell-side or by third parties.


MiFID II’s emphasis on best execution stems from a focus on investor protection. In the case of one segment of the buy-side, corporate treasury desks, that is often less under scrutiny because many of them reputedly do not see their roles as competing with other desks but rather reducing risk within given sets of parameters. In other words, their fiduciary objectives are often not geared necessarily towards best execution but rather towards risk reduction.

But even for this segment, Lee reports increased appetite for algo usage. He said that from the conversations he’s had during the past 12 months, he was personally a little surprised at the level of interest by the increasing number of corporates that are using FX algos to hedge. Lee did add that one factor that may have had a hand in this was the Swiss National Bank policy shift early in 2015, which underlined for corporates how much an FX move could affect their bottom lines.

“I think that that closer link between people and their algorithms is of huge benefit for the market. It means the sensible and safe use of algorithms,”

An additional encouragement comes from the degree to which algos can reduce risk. Cochrane pointed out that with a voice trade, beyond the currency risk, there is operational risk, clearing risk and confirmation risk, among other factors. “And an algo gets rid of a lot of the other risk. It’s more efficient.”

If an algo reduces all of those risks and helps a firm show that it is getting best execution, why wouldn’t someone want to use one? The key, according to Cochrane, is that trust needs to be built.

“People don’t understand the inner workings of the algorithms so they’re nervous about using them, regardless of whether it’s a bank or an agency broker, because they’re afraid to go to their board of directors and get the hard question: what do you know about that algo?”

ITG has heard from some of its larger accounts that they don’t want to use an algorithm unless they understand how it fills the trade down to the minor detail. And that, he said, takes time. “They have to really understand every little detail about the algo, where the rates are coming from, is there last look, no last look pricing, is there any ability whatsoever to get front-run on a position?”

Going forward, he said the answer has to be in education and increased transparency. He sees it as practically a requirement for banks, and to some extent brokers, to explain to non-adopters or slow adopters exactly how the algorithms and how the agency model work.

Showing clients where the market was and where they executed – Execution Quality Analysis – and how each algorithm works will be needed to increase the confidence. “Then I think it will become a more natural progression to where the clients who are more cost conscious and are using algorithms and are using them to great effect will start to have a competitive edge,” Cochrane said.

At that point, when market participants see that algo-using segments of the market are outperforming, it will snowball.