The Importance of Trust in Cross-Border FX

Trust is important when you're moving millions around the world.

Every cross-border FX trade begins with two parties who are definitionally interested in gaining an advantage over one another. The buyer wants a better price, the seller wants to protect their margin. Neither side has complete information.

These are not conditions under which trust naturally springs.

In deep markets, much of the consequences of this situation are absorbed by the book itself. The market provides a list of prices and counterparties, the efficiency of the market means that those prices remain predictable except under the conditions of extreme strain.

If something seems strange, there is enough information to make a decision with confidence.

In smaller FX markets, which notably make up most corridors on the planet, the situation is far murkier.

In many of these markets, trading still largely happens through chat or RFQs rather than against an electronic book. The quotes offered are not necessarily “real” in the sense that they can always be executed on, and under last look a provider can still reject the trade entirely even after you accept their quoted price.

Worse still, the information available to you is often terrible, particularly when we are talking about liquidity depth. A provider that might be able to quickly and cheaply move $100,000 USD to BRL, might fall apart completely at $5 million. It may be impossible to tell that this is the case until you've already requested a quote and seen your normal spread widen 20 bps.

This is why most treasurers don't put all their eggs in a single basket, relying instead on multiple liquidity providers and hoping that by "averaging" them out, they can reduce their overall execution risk.

If a provider is proven trustworthy, they move more flows. If they are not, they may be cut out entirely.

Three states of trust

This offers a useful way to think about what "trust" means in a market like this, and the relationship it creates between a liquidity provider and their customers.

I see trust as existing in one of three states, each determining how a customer treats their next trade.

• No Trust: The customer has very little evidence about how the provider behaves, so they test them. They may trade smaller clips, or be more prone to splitting their flows against competitive providers. For more sophisticated teams, the results of these tests might go into aggregate risk scores that drive algorithmic trading systems. Every transaction helps to determine whether or not the next transaction will occur.

• Trusted: The customer has enough history to make reasonable assumptions about what will happen next. They understand where the provider is strong and what normal execution looks like. In this phase, that reliability becomes, to a greater or lesser degree, as important as small variations in price or settlement time. Additional flows and less competitive search behavior often follows.

• Trust Lost: Something has happened that makes previous assumptions indefensible, the provider can no longer be counted on to offer the service they claim at the level of reliability the customer has come to expect. This is a far worse condition than "No Trust," because the customer now has evidence of failure rather than simple suspicion.

Customers move between these states based on the evidence they collect from every transaction they make. Better outcomes push them towards trust, while worse outcomes move them to the edges. This raises the question, what does a "good outcome" actually look like?

Reliability and predictability

A "good outcome" consists of providing the customer with what they expect: Request a quote, confirm a price, fill at the speed that was promised. Do it again. We write a lot about "high quality liquidity," and this is what that means in practice. Over enough trades, the customer learns that the provider is reliable, and feels more comfortable giving them more business.

Reliability is critical, but it's not the only factor at play. A technically reliable provider can fail at the tail. Imagine one firm that settles 90% of its transactions in around 30 seconds, with the remainder taking between four minutes to three hours. Another normally takes 90 seconds, but almost every transaction arrives inside two minutes of that. The first provider is much faster most of the time, but the second is considerably easier to build a business around.

The difference is practical.

A treasury team needs to be able to predict when a transaction will land if they ever hope to build automations to support their trades. The less volatility exists, the easier these predictions are, and the more smoothly their business runs. If there is a chance, even a small chance, that a transaction will take five or ten times longer than expected, they must plan as if that is the actual constraint.

Reliability tells you how often the provider delivers. Predictability is determined at the tail, it tells you how far the result is likely to wander from what you have learned to expect.

Volatility is the enemy of both.

Trust is earned by reducing that volatility.

How trust changes flow

The same transaction can mean very different things depending on where in the process a customer happens to be. When the customer is still early in the relationship, a successful trade teaches them just a little bit more about what the provider can do. After a few dozen trades like that, those lessons become closer to proof.

And proof drives volume, once a customer believes in the provider's execution quality, they may start integrating their systems more deeply through an API or other automations.

Shopping behaviours decrease as trust ramps up.

The cost of failure at this point can be enormous.

A bad trade from an unfamiliar provider confirms very little because the customer is still learning. The same failure from a trusted provider can force the customer to question the entire relationship. This is made far worse if the provider fails to give good answers for their failures. Trust is often as much a customer service question as a systems one.

Trust is earned slowly and lost all at once. Once we understand that, the next question is straightforward: what does a provider have to keep getting right for those assumptions to survive?

What trust is worth

Everything.

In an adversarial market, customers are constantly building a model of their counterparties to get a sense for how far they should be believed. They need to know not only what normal trades look like, but where the boundaries sit. They also need to feel confident that if something goes seriously wrong, their provider will be there to make it right.

Cross-border FX is not only an adversarial market, but one where information is uniquely difficult to find. Whether you are talking about price or depth, a treasurer has no "objective" source they can turn to other than their counterparties. The smaller the market, the fewer the sources, and the more opaque the information that remains.

Trust is everything because it allows the treasurer something they prize, legibility, and legibility allows you to make decisions.

Once customers can predict what will happen next, they have a reason to come back.