New FX technology resolves phantom liquidity
Fragmentation and phantom liquidity bedevil the foreign exchange market with its proliferation of trading venues. Investors and corporates want to see the true depth of the market and what amounts they can really trade at the enticing prices being flashed at them. In a low-return world, these end-users are getting rigorous on trading cost analysis. Banks are developing new technology to respond.
How many electronic trading venues are there for foreign exchange, including all the retail aggregators, all the multi-bank platforms aimed at different customer groups, specializing in different combinations of spot and FX derivatives? Euromoney asks each head of foreign exchange at the leading banks and gets different answers. But it’s a lot: anywhere from 40 to 50 in total, of which at least 10 and maybe as many as 20 are quite important to the wholesale banks.
David Fotheringhame, director at Barclays, reflects on how the barriers to entry in the foreign exchange ECN business have come right down. “When EBS and Reuters were being built, it was like building the space shuttle,” he says. “Today, it’s a lot easier to get the technology in place to set up a new ECN and attract some reflected liquidity onto it.
The rules of the leading trading venues, their choices as to whether or not to display volumes, are far from uniform. They also change regularly, which in turn impacts which users put volume through the venues, and the depth of markets and reliability of the pricing they show.
This is aggravating for end-users who, given their own regulatory requirements, operational risk budgets and tightness of IT resources, would much prefer a uniformity of trade processes.
The biggest problem is phantom liquidity, as many venues display tight pricing while relying on just a handful of large market-making banks to provide it.
“In FX liquidity, it’s all down to the firmness of quotes published and whether the price you see is even close to the price you can get in meaningful size,” says Fotheringhame. Last-look rules in certain venues allow them to reject orders even when participants have hit an apparently valid displayed price.
“A lot of the apparent liquidity is actually spurious,” says Fotheringhame. “That makes it very difficult for customers to achieve optimal execution in the FX market. Just because you see a tight price in $20 million on one venue and $20 million on another doesn’t mean you’ll actually be able to get $40 million done at that price. You might not even get $10 million done.”
Barclays has been a pioneer in electronic foreign exchange with its Barx system. Its latest upgrade to this is called Gator. It was designed as an aggregator across the various venues for the bank’s own traders, built on algorithms developed from constantly submitting small and large orders across all the venues at various times of the day, testing the true depth of market and reliability of price.
It now allows Barclays to make a very good estimate – not it insists, an absolute guarantee – of the best actual price a client could fill a certain sized trade at, splitting it across the various venues. It intelligently gauges that actual depth of the market.
The calculations behind this are not static. Barclays refines them as it continues to submit orders and execute trades across all the venues.
The bank has decided it should roll this out to clients, on a fixed-fee basis, charging a commission per trade. This is built into the customers’ spread rather than paid later as a separate cheque.
Mike Bagguley, head of foreign exchange and commodities at Barclays, says: “We operate in a market where customers can demand the tightest pricing but also top-quality service.
“The solution is certainly to be a volume player but also to run a business with a high intellectual property component, whether it be in relative value trade ideas, insight around trade flows or IT related, as our new aggregator system is.
“If Barx 1.0 was streaming two-way prices and Barx 2.0 was the order book, Gator is Barx 3.0. It is a profound change to the marketplace and one that should sustain us for the next two or three years.”
As well as doing the job for clients of second-guessing actual price and depth of market versus that being displayed across all the many venues, Gator aims to reduce operational complexity.
If a client submits a $20 million order through the system, that might result in multiple trades across various venues that finally leaves it with an unfilled stub of say $0.5 million. Barclays will complete that as principal to ensure the fill and deliver the multiple underlying trades as one order.
Fotheringhame says: “We call it a hybrid liquidity model, consisting mainly of external liquidity that we execute against on an agency basis but with some principal as well.”
Fotheringhame politely declines to discuss any of the internal debate that prevailed before Barclays took this rather impressive-looking tool for its own traders and offered it to customers. Obviously the commission revenue, though tiny per trade, helps it recoup some of the substantial costs of development.
“We really are as much a technology company as a bank now,” says Fotheringhame. “We’ve had around 100 people working on this which we can justify as a leading wholesale foreign exchange bank. Not many customers can afford those teams of developers though. It was expensive to build this.”
A cynic might also note that, when the internal budget arguments next prevail inside Barclays over IT resources, the foreign exchange business will be able to say that it really does need that next generation update on Gator not just for its own traders but for all the customers it signed on to it.
However, that’s standard bank budget politics. What does the system do for customers?
Fotheringhame slaloms through a few test runs, asking the system how much of a $100 million order in a leading currency pair Gator thinks could get done at the mid point of the best displayed price. The rather disappointing answer is just $16 million.
However, he also shows how close to that best price the customer could achieve if it could speedily submit and execute orders across many venues. Possibilities even open up to deal at better prices than had first appeared. The order can get done – but it will comprise a whole raft of smaller orders.
It’s with more complex cross-currency trades – in currency pairs that require two or more legs to achieve, for example a euro-yen trade involving euro-dollar and dollar-yen legs – that the usefulness of the system for taking the complexity out of the process for customers reveals itself.
The Gator system shows a euro/yen market in fact derived from pricing in two separate underlying currency pairs taken from multiple venues.
Fotheringhame takes a more extreme example, and does a one million Aussie dollar/Swedish krona trade. It amuses Euromoney to see him calmly throw this into the system to demonstrate its capacity: a tiny drop in the vast ocean of foreign exchange volume.
That order is filled in about one-and-a-half seconds, but when Fotheringhame reveals the underlying trades that took place to fill it, including a small stub piece taken as principal by Barclays to be internalized, there are more than 15 individual transactions across three underlying currency pairs, each sending their own confirmation messages rippling around the global market place waiting for settlement.
Fotheringhame makes no bones about Gator’s popularity with customers. “They’re biting my arm off for this,” he says. “Taking out the stub amounts, for example, removes remaining odd-lot trades in currency they don’t want that they would otherwise still have to monitor and settle.
“This takes out a lot of the back-office and middle-office complexity, as well as delivering an intelligent trading system.”
All around the foreign exchange markets, it’s the same story: the key issue for banks to wrestle with is IT resources devoted to systems maintenance, new developments such as Gator and even Tradition’s ParFX, which several banks are funding (see Euromoney May 2013 banking news section) as well as other, smaller new applications.
The economics of the business are changing from a pure principal trading one to incorporate a larger component of commission earnings, such as the ECNs charge and aggregators such as Gator.
|Eric Auld, global head of FX and FX hybrid trading at BNP Paribas|
Eric Auld, global head of FX and FX hybrid trading at BNP Paribas, says: “One of the biggest questions we are asking ourselves is how client behaviour evolves as the operating model of the FX business tends more and more towards the equity business model.
"Whether we get to a fully commission-based business model is open to debate, but the direction of travel towards that state is clear.
"And as the “fees” that clients pay to their counterparties for conducting their foreign exchange business become more transparent because of that direction of travel, so clients become more rigorous about how they allocate their business to reward the perceived value they receive.”
|Pete Eggleston, head of Morgan Stanley’s quantitative solutions and innovations (QSI) group|
Morgan Stanley is testing the ground for a consultancy-like advisory service in trade execution. Pete Eggleston runs the firm’s quantitative solutions and innovations group, which has been devoting a lot of attention to post-trade analysis of foreign exchange transaction costs, looking beyond simple best execution against mid-market and more at the potential for customers to execute trades that don’t move the market price against them.
This will only become more important to investors as they increase exposure to foreign exchange, and also as liquidity fragments and becomes more variable.
Eggleston says: “We are in dialogue with several hundred large investors including real money, hedge funds and sovereign wealth funds, for many of whom a key theme is diversifying away from traditional and at times highly correlated asset classes into, among others, foreign exchange as a source of return.
“And we have various methods to help investors do that, from a platform for them to manage specialist FX managers, to a structured FX product with exposure to multiple FX strategies and styles.”
This is all the pre-trade, big picture stuff: whether to devote risk budget to FX at all, whether to have internal managers run FX exposure or to hand it to external managers or buy a structured exposure.
Next on clients’ to-worry-about list is transaction cost analysis. Regulators are all over this. If institutional investors are going to charge execution costs to end-customers, what do they consist of and how, in a highly competitive low-return marketplace, can they be kept down?
Eggleston says: “This is very cutting-edge, 21st century FX trading, looking at pre-trade aspects, then offering various execution mechanisms – voice, electronic or algorithmic – to post-trade, providing a transaction cost analysis.
“For that analysis to be relevant you really have to understand the micro-structure of the FX markets and see that best execution is more than just best price at any given moment, but must also consider market impact and slippage, and the trade-offs between cost and speed and certainty of achieving an order.
“We’ve done VWAP [volume-weighted average price] for FX and it’s quite different from VWAP for equities, partly because foreign exchange is an over-the-counter market.”
How does this work in practice? In certain currency pairs, it can be surprisingly basic.
Eggleston says: “While FX is generally held to be robust and highly liquid, there are certainly markets where investors can have impact and hurt their own execution even with relatively small tickets.
“It’s actually quite an eye-opener to some customers’ central trading desks that it might be better to execute orders in central and eastern European currencies at 3pm London time, rather than gather them all up through the day and execute them together at 6pm when liquidity is much diminished.
“Or we might advise an M&A client with, say, a couple of yards of Aussie to execute, on working the order in stages over a period of time rather than quickly at the risk of potential slippage.”
And this analysis isn’t based on just Morgan Stanley volumes and price history. “We can base it on a big percentage of electronic volume data from a number of sources,” says Eggleston. “We’re developing this to look through phantom liquidity and analyze how best to size and time trades, and allocate them between venues to minimize market impact.”
It’s a slightly different take on what Barclays is doing with Gator. Eggleston says: “I was with a large institutional investor recently who had worked out that by adapting his execution style using this advice, he could save in the region of 30 basis points. Based on the annual volumes this investor does, it’s a very big sum of money.”
Is this obsession with transaction cost analysis at the big foreign exchange banks an example of the turkeys voting for Christmas? One head of FX thinks so: “Clients will happily take tools off you to minimize their trading costs, but they won’t pay you for them.”
While Morgan Stanley likes to call its work with clients on execution an advisory service, it gets paid in FX volume put through its traders not in cash fees.
Regulators’ obsession with best execution lies behind the growth in this kind of analysis of the spot foreign exchange markets. “It’s very difficult to know for sure, if you’re a customer, that you’re getting best execution, but all banks are certainly trying to demonstrate it,” says the head of G10 FX trading at one leading bank.
“We’ll do a fully audited performance check for customers to compare against EBS or Currenex.”
The impact of regulation hasn’t hit foreign exchange quite as hard as the banks’ other FICC businesses, notably credit, but the banks are all worried that it could put a sizeable dent particularly in the FX derivatives business.