Forex ‘last look’: how non-banks stack up

Research shows patchy disclosures, plus differences from banks on pre-hedging and rejected orders

Risk 0819 In depth Mark Long nb illustration
Mark Long nbillustration.co.uk

  • While non-bank liquidity providers are becoming important players within the spot forex market, they disclose less about their dealing practices publicly than dealer banks.
  • Risk.net combed public disclosures and gathered non-public information on six non-banks and five of the largest forex dealers.
  • The resulting table compares controversial areas such as approaches to pre-hedging, application of hold times and use of rejected order information.
  • Just three of the top six non-bank liquidity providers have public disclosures, compared with all five of the top dealers.
  • This is important, as end-users can have indirect exposure to non-banks’ dealing policies.
  • Some non-banks stick out for their use of pre-hedging in the last look window.
  • “It’s our duty to police these counterparties and make sure they’re doing what they should be doing,” says James Binny, currency head at State Street Global Advisors.

This is the first of a two-part series looking at last look disclosures in the foreign exchange spot market. Part two looks at disclosures of the top 50 liquidity providers and is available here.

While becoming increasingly important players within the electronic foreign exchange cash market, some non-banks are not necessarily playing by the same rules as their bank competitors. Non-bank liquidity providers have all adhered to the global forex code, but compared to the top five banks, their trading practices can vary widely – if they are disclosed at all.

The differences lie in how they treat client orders once a rate has been agreed in the so-called ‘last look’ period, where the dealer will conduct final checks – surrounding the client’s credit line, the details of the trade, the current market price, and sometimes with an add-on for further monitoring – before deciding to accept or reject the trade request. Liquidity providers all have different policies around how they use that order information, and the circumstances under which they reject trades.

The code – originally published by the Global Foreign Exchange Committee (GFXC) in May 2017 following a series of fixing scandals – sets out conduct standards for trading in the forex market. Part of the code asks liquidity providers to disclose their last look policies.

When Risk.net collated last look practices from dealers last year, it focused on the top 15 banks. But recent surveys, such as the annual Euromoney forex poll, shows market share has been growing for the non-banks. While some believe such surveys should be taken with a pinch of salt, it is clear these types of firms are becoming increasingly present in the market, inviting closer scrutiny of their last look practices, and how they compare with their bank rivals.

This year, Risk.net combed the public disclosures of five of the biggest forex dealers and the top six non-bank liquidity providers. While all the dealers provide public last look disclosures, only half of the non-banks follow suit – and many with only partial disclosures – making their policies in this often controversial area of the forex market rather opaque. This is important, as some banks’ execution styles give end-clients indirect exposure to non-bank last look policies.

The non-banks that did not publicly disclose their last look policies explained their approaches separately to Risk. The results are in the tables below.

Putting the public and non-public disclosures together, it shows a spread of last look approaches from the non-banks. While some are proverbial boy scouts on issues such as hedging in the last look window, others are either silent, or engage in what some describe as potentially “predatory” practices.

“It’s interesting that the non-banks seem to be coming out quite badly on this,” says James Binny, global head of currency at State Street Global Advisors (SSGA). “I would have said they’re introducing extra competition to the market and that’s a good thing because that should ultimately mean we get a better service for our clients, but I find it a bit disappointing that they’re coming out quite badly here in terms of transparency. It seems a bit strange.”

But James Wood-Collins, chief executive officer of Record Currency Management, cautions not to paint all non-bank liquidity providers with the same brush.

“Within the non-bank market-making community, there is a range of behaviour from those who genuinely see their function as market-makers and want to be committed to legitimate risk management, and then those who perhaps take a stance which is rather more predatory. I think it’s important to recognise that not all liquidity providers are the same by any means – and that’s true of banks too,” he says.

Price check and hold time

Market participants each have their own definitions of what last look consists of, but broadly it consists of two elements. The first is a price check conducted by the dealer – which could last anywhere from five to 100 milliseconds – where they can see multiple price updates from trading platforms. Depending on the dealer’s policies, it can then decide to reject an order if it moves too far from an agreed price.

The second element is a so-called hold time, or speed bump, which can be applied on top of the last look check – and could be used to observe if a client’s behaviour leads to market movements against the liquidity provider, making it harder to lock in a profit. Critics say that if pre-trade checks are complete, the only reason to slow down execution is to allow the dealer to see more price movements, which could result in more trade rejections if the price moves away from the bank, and for those using symmetrical approaches, profits if it moves toward the bank.

The global forex code asks adherents to disclose their last look policies to clients. It does not specify that this needs to be a publicly accessible document, but a February 2019 report from the Global Foreign Exchange Committee suggested public disclosure was one way to improve transparency.

James Wood Collins of Record Currency Management
James Wood-Collins

While all five of the top banks disclose their last look policies publicly, only three non-banks – HC Tech, Virtu and XTX – do the same.

Given their main distribution outlet has been anonymous electronic communication networks (ECNs) such as EBS, which serve the interdealer market, it is easy to dismiss the impact of a lack of disclosure by these firms on end user clients.

However, five of the six non-banks confirm they have client businesses, where they directly provide forex prices to users via direct connections or platforms.

But end-users can be exposed to non-bank last look policies indirectly as well. For instance, a regional bank that sources its liquidity for clients via a direct stream from a non-bank, will create an exposure if that bank takes a so-called cover-and-deal approach to trading.

This is where they source the hedge from non-banks before executing the trade with the client to avoid taking market risk. That means a client’s order information is transmitted to the non-bank liquidity provider before the trade is executed, which indirectly creates an exposure to the non-bank’s last look policy, of which only half are publicly accessible.

A similar situation arises if a bank employing cover-and-deal sources liquidity from a non-bank via an ECN. The last look policies of ECNs are also opaque, with two leading platforms confirming that they do not have specific last look standards in place – again giving clients indirect exposure to non-banks’ trading standards.

Differing approaches

When public and non-public information on last look policies is gathered up, it is clear there is a spread of approaches from the non-banks.

“We trade with a large number of non-banks, and some of them have really been at the front of the industry compared to banks in terms of how they disclose their own practices. [But] there’s a lot of variation within the non-bank community,” says Christian Lønborg Thomsen, forex liquidity manager at Saxo Bank.

This can be seen in one of the most controversial areas of last look, which is whether firms pre-hedge their clients’ trades during the last look window. Principle 17 of the code advises market participants against trading activity that utilises the information from the client’s trade request during the last look window, including hedging.

“Such activity would risk signalling to other market participants the client’s trading intent and could move market prices against the client. In the event that the client’s trade requests were subsequently rejected, such trading activity could disadvantage the client,” the code says.

The code provides an exception if firms are taking a cover-and-deal approach, and clearly communicate this in advance. However pre-hedging generally makes some clients uneasy: “Pre-hedging during the last look window is undesirable, because liquidity providers are starting to know what you’re doing at that point and moving the market against you,” says SSGA’s Binny.

We trade with a large number of non-banks, and some of them have really been at the front of the industry compared to banks in terms of how they disclose their own practices. [But] there’s a lot of variation within the non-bank community
Christian Lønborg Thomsen, Saxo Bank

All five of the dealers clearly state in their public disclosures that they do not pre-hedge in the window, but non-banks are less consistent. Only one – XTX – publicly states in its disclosures that it does not pre-hedge in the window. When asked, two more – Citadel Securities and Flow Traders – confirm they also do not engage in the practice.

Meanwhile, HC Tech states in its public disclosure that it will utilise pre-hedging “as permitted by the FX Global Code” and therefore may pre-hedge counterparty trade requests when acting as principal.

The disclosure says the firm “engages in pre-hedging with the intent to advantage the counterparty with optimal fill rates, but also may receive a benefit from pre-hedging the transaction”. It also notes that its pre-hedging may result in an “overhedge” of the amount initially requested, but that it fully accepts the risks and doesn’t use the strategy if the firm believes it will disadvantage a client’s trade request.

A spokesperson for HC Tech confirms that it does not pre-hedge direct client flow. Pre-hedging is done either with the consent of the client, or after notification.

Virtu’s dealing disclosure is a little more vague, noting that it may pre-hedge “prior to or alongside a counterparty’s transaction” and may even trade in a way that “adversely affects or is otherwise not consistent with the interests of its counterparty”. But it emphasises separately that it would only pre-hedge in the window if clients agreed.

Jump confirms to Risk.net that it does pre-hedge in the window, but only for a specific product, where it converts futures liquidity into cash for its counterparty, and with counterparty consent. “The process is transparent and documentation is in place,” says a spokesperson for Jump.

Breach policies

Another controversial aspect of last look is how firms treat excessive price movements within the window. While some liquidity providers’ approaches are very nuanced, they essentially have two broad choices: a symmetrical approach, where moves for and against the client are rejected if they go beyond a given threshold during the last look or hold time window; or an asymmetrical approach, where only moves that benefit a client are rejected if they breach a threshold.

While on the surface the asymmetrical approach seems unfair, some dealers have argued it provides better pricing for the client.

All five dealers disclose their approach. Four of the five use symmetrical, with Deutsche Bank taking a default asymmetrical approach and giving clients the option to switch to symmetrical instead.

Once again, non-bank disclosures in this area are patchy. Only two firms disclose their approach publicly: XTX takes a symmetrical approach, while HC Tech uses asymmetrical, but like Deutsche allows clients to switch if they desire.

My guess is that if a liquidity provider doesn’t come out and trumpet the fact that they’ve got good features about their liquidity provision, then they probably don’t have them
Kevin Rodgers, consultant and former Deutsche Bank forex head

Separately, Jump, Flow Traders and Citadel say they follow HC Tech’s approach.

Hold times are the other area that can differ significantly between firms. On the dealers’ side, four of the five banks confirm their policies in their public disclosures. UBS and Deutsche apply hold times, while JP Morgan and HSBC do not. Citi declined to comment.

Only one non-bank – XTX – publicly discloses that it does not apply a hold time. Flow Traders told Risk.net that it also has no additional hold time. Citadel says hold times are customisable. Jump says hold times are up for discussion with clients, and Virtu says the hold time varies by client.

One area where banks and non-banks are aligned though is how they disclose the use of potentially lucrative information from orders they have rejected. The global code doesn’t explicitly require this, but Guy Debelle, deputy governor of the Reserve Bank of Australia (RBA) indicates it might at some point. Only a handful of liquidity providers currently disclose their policies (see box: Use of rejected order information).

Decent disclosure?

With so much focus on last look practices in recent years, the question is why non-banks would not want to disclose publicly.

Consultant and former head of forex at Deutsche Bank, Kevin Rodgers, says if these firms had a good story to tell, they’d be shouting from the rooftops.

“My guess is that if a liquidity provider doesn’t come out and trumpet the fact that they’ve got good features about their liquidity provision, then they probably don’t have them. So in my opinion, whatever tick-list of good stuff that you want about last look, if people aren’t mentioning it, then they probably aren’t offering it,” he says.

Mark Bruce, head of fixed income, currencies and commodities at Jump Trading, defends his firm’s approach, saying public disclosures can actually be more of a hindrance than a help to clients given the confusing legal language they contain.

“We have found that disclosures varied in the way they’re written and the products they cover,” says Bruce. “Without an explanation of any nuances, they can cause quite a lot of confusion for counterparties. Our preference is to share our disclosure with counterparties directly, going through everything in detail and customising the way in which we trade with them.”

So just because market participants have signed the code it doesn’t necessarily follow that those people are well behaved in deed and act
Darryl Hooker, Harperdan Consulting

Given the divergence of last look practices and language contained within disclosures, SSGA’s Binny highlights that the buy side ultimately has a large role to play when it comes to the surveillance of last look practices.

“I think a lot of what went wrong pre-crisis was that the buy side were not policing the sell side hard enough and this all comes into it. It’s our duty to police these counterparties and make sure they’re doing what they should be doing,” he says.

Darryl Hooker, chief executive officer of Harperdan Consulting, agrees counterparties should engage in greater surveillance of their liquidity providers, as a commitment to the global forex code doesn’t automatically mean liquidity providers are engaging in best practices within the forex market.

“Because we have policemen, do we all cross the road at zebra crossings? The simple fact is no, we don’t. So just because market participants have signed the code it doesn’t necessarily follow that those people are well behaved in deed and act – the client needs to dig deeper and look at the disclosures they [may] have been given to see if they clearly understand the basis upon which their transactions requests are being accommodated. That’s where the question lies,” says Hooker.

Use of rejected order information

While sitting outside the last look disclosures, one issue that is increasingly gaining attention is the extent to which liquidity providers use information from foreign exchange orders they have rejected, for trading or risk management purposes.

Although not explicitly mentioned within the forex global code in its own right, in its February report, the GFXC provided additional guidance over market participants’ use of such information, suggesting it as a possible topic of conversation in bilateral meetings between liquidity providers and clients.

Disclosures on the practice are patchy, from both the bank and non-bank liquidity providers. But Risk found that only one non-bank explicitly states within their disclosure that they do not use rejected trade order information: XTX Markets. Five more confirm they do not use this information – Citadel, Flow Traders, HC Tech, Jump Trading and Virtu.

Some market participants worry that rejected trade order information could be used in a predatory fashion – if a liquidity provider knows a client will try and trade again soon after the rejection, it knows which way the market will move in advance, potentially leading to profits for them and a worse rate for the rejected client.

“Use of rejected order information is a huge concern. What valid reason could there be?” asks Matt Clarke, Europe, Middle East and Africa head of distribution and liquidity management at XTX Markets.

He says a liquidity provider that uses rejected order information can ultimately increase the market impact of a client’s order, leading to higher transaction costs.

“A client recently shared some data which showed that the market impact on their rejected orders was eight times that of their filled orders. The cost of these rejections from each liquidity provider and venue is worth monitoring closely,” he adds.

A 2017 report by LMAX Exchange shows a similar market impact in regards to rejected trade orders, where a trader receiving a 5% reject rate and a 100 millisecond hold time is estimated to experience an extra trading cost of around $25 per million in aggregate over their total trading volume.

Surprisingly, on this point the top five banks don’t fare much better, with only Deutsche Bank and HSBC stating within their disclosures that rejected order information is not used. When contacted, UBS and JP Morgan say they do not use rejected order information. Citi declined to comment.

Some clients see the use of rejected order information as a violation of the spirit of the forex global code.

“To me, using rejected order information feels like it could be a predatory use of last look rather than just a risk management use,” says Wood-Collins of Record Currency Management. “I don’t know whether you can point to something in the code which explicitly prohibits that, but it certainly feels like that goes against the intention and purpose of the Code.”

Guy Debelle
Photo: Reserve Bank of Australia
Guy Debelle

However, Debelle of the RBA and chair of the GFXC, says the use of rejected order information isn’t such a black and white issue.

“I don’t think there’s always an easy answer. I can think of examples where there’s not an obviously right or wrong answer,” he says.

Binny at SSGA agrees, highlighting that while there are some cases where the use of such information could genuinely be for risk management purposes, there are others where the use of such information verges on being exploitative.

“For genuine market-makers who are trying to hedge their books, I can understand why that information is useful. I think where it becomes dodgy is where people are quoting without any real desire to win the business. People should make markets to make money from making markets, and so should be making money from the spread, not from exploiting the information you’ve just given them,” he says.

Debelle says that while disclosures should mention if rejected trade order information is used, there is no principle in the code or provision within the GFXC’s February report which prohibits the use of such information. As a result, clients should ask their liquidity providers if and how that information is going to be used and trade accordingly.

“If you don’t like the information from within the disclosure you’re getting, then don’t deal with that person. That is always an option available to you,” he says.

That being said, Debelle reiterates that market participants should consider stating the use of rejected trade order information within their disclosures and confirmed that the GFXC may look to provide more guidance in this area, such as through examples annexed to the code.

But Hooker of Harperdan Consulting says this may not be a silver bullet: “We could all be writing examples for the rest of our lives because everyone wants more examples. I think in certain circumstances more examples would help definitely, but whether that gives us more clarity I’m not sure.”

Additional reporting by Robert Mackenzie Smith

Update, August 2, 2019: This article has been updated with new information about Flow Traders’ dealing practices. The firm gives clients the option of switching to a symmetrical approach to rejecting trades that move excessively during the last look or hold time window. The firm does not use rejected trade order information for trading or risk management. The original information Flow Traders provided to Risk.net was inaccurate.

Only users who have a paid subscription or are part of a corporate subscription are able to print or copy content.

To access these options, along with all other subscription benefits, please contact info@risk.net or view our subscription options here: http://subscriptions.risk.net/subscribe

You are currently unable to copy this content. Please contact info@risk.net to find out more.

Most read articles loading...

You need to sign in to use this feature. If you don’t have a Risk.net account, please register for a trial.

Sign in
You are currently on corporate access.

To use this feature you will need an individual account. If you have one already please sign in.

Sign in.

Alternatively you can request an individual account here