Central banks home in on FX trade execution

All the signs are that foreign exchange (FX) market participants, ranging from corporates to asset managers and central banks, are going to be using more sophisticated FX execution techniques, coupled with detailed transaction cost analysis (TCA), in the coming 12 months, as those who have lagged behind catch up with their peer groups.

Although regulatory initiatives have so far largely not been directed at spot FX trading, there is a general move towards more transparency right across the financial services sector. And lawsuits in the US involving pension funds suing banks for overcharging on their FX trades have helped put the spotlight on best execution.

Technology and manpower spending among the top 10 global FX banks have been running at very high levels in recent years, as they roll out new and more sophisticated algorithmic programmes designed to execute large trades by searching all possible venues for pools of FX liquidity. With spreads on highly liquid currency pairs such as euro/dollar having come down to inside three basis points for the largest clients, FX banks need to generate more revenues from service packages and less from traditional spread retention.

That will include advisory to corporates on how to execute relatively small trades in illiquid currencies, as well as providing increasingly powerful algo execution tools to large FX players such as asset managers and central banks.

Reserve diversification

Central banks are increasingly active players that require trades in very large sizes. Adrian Boehler, global head of institutional FX sales at BNP Paribas, says central banks tend to act like a traditional asset manager due to the sheer size of their reserves.

"Central banks consume the same sort of FX services as real money investors such as pension funds, insurance companies and asset managers. Those services can include post-trade transaction cost analysis, and will increasingly include pre-trade analysis on how best to execute an order at any given time," says Mr Boehler.

Central banks have typically been more conservative in their approach to electronic execution than other institutional clients, but Mr Boehler believes algorithmic execution can provide them with the benefit of being able to slice up large orders and minimise their footprint in the market. Most of the algorithmic execution at the moment takes place in G-10 currencies, but BNP Paribas expects that demand will keep on growing for algorithmic execution in emerging market currencies.

Kiran Kowshik, currency strategist at BNP Paribas in London, says the bank has a bearish view on the dollar that is partly related to the fact that central bank reserves, and hence global US dollar supply, are still growing despite fears of a Greek eurozone exit and a hard landing in China. Reserve growth is not collapsing as was the case in 2008 and 2009, as central banks are maintaining an active policy aimed at providing ample global liquidity.

"Given that central banks mostly receive inflows in dollars, and then have to make allocation decisions, we believe that the general trend of diversifying away from the dollar will continue," he says. "So although some of our competitors are predicting a slowdown of central banks as consumers of FX liquidity, we take a different view," says Mr Kowshik.

Catching up with equities

The FX sector has lagged behind equities in areas such as algo execution. But FX presents different challenges, for example, in placing a participation cap in order to avoid moving the market with a large trade. This is a relatively simple matter in equities trading, where on-exchange volume trends are easy to identify, and a calculation can be quickly made as to whether a buy or sell order will move the market. But the over-the-counter nature of FX trading, and the proliferation of bank platforms and independent electronic platforms such as FXall, make assessing trading volume a difficult task.

There is a balance to be struck between accessing liquidity from as many sources as possible, while avoiding information leakage and providing sophisticated investors such as currency hedge funds with signals to trade on. Accessing the deepest possible liquidity with a minimum market footprint is the goal.

"In the algo area, time slicing is of key importance, since liquidity can fluctuate during the course of the day," says Pete Eggleston, head of the quantitative solutions group at Morgan Stanley in London. "There is a widespread perception that a currency pair such as euro/dollar is so deep and liquid that it is hard to impact the market, but our models clearly show that at certain times of the day it is possible to create quite significant market impact with relatively small trades."

The bank’s Matrix execution platform offers algo strategies such as MS Fix – a volume weighted average price (VWAP) algo that allocates higher proportions of a trade to more liquid times of day, thereby helping to minimise the market impact of a large trade. And the Matrix platform will soon launch MS Radar, where the client sets out parameters such as maximum bid/offer spreads. The algo will then opportunistically execute trades within those parameters, automatically pausing and restarting as market conditions change. All FX executions done via the Matrix platform come with a full range of pre-trade and post-trade TCA.

"Our focus is on advising clients to become more efficient in their FX execution processes, and the solution may or may not include algorithmic trading," Mr Eggleston explains. For example, a client might be executing trades in central and eastern European currencies after the close of trading in London, and simply by moving those trades to earlier in the afternoon can access better liquidity and save on spread costs.

"Whether they are trading FX as an asset class, or as an ancillary process to equities and fixed-income trading, investors such as pension funds and asset managers are looking much more closely at the quality of their FX execution than they were three or four years ago," says Robert Maher, global head of fixed-income electronic sales at Credit Suisse in London.

Learning from the past

Credit Suisse's Advanced Execution Services (AES) platform was one of the first movers in providing best execution services in FX, building out from its origin as an equities product. Mr Maher says AES now has clients right across the spectrum, with a split of about 60:40 between the G-10 developed market currencies and the rest.

“It could be a client conducting relatively small trades in illiquid currencies, or a major corporate or real money account trading large tickets in liquid pairs. In both cases, the trades need to be broken down by algos into many smaller anonymous trades, so as not to send out signals that move the market," he says.

In 2011, Credit Suisse launched Instant Colour, a TCA product giving clients real-time information as a transaction progresses, and immediate analysis on the efficiency of the trade once it is completed. This year a pre-trade analysis function has been added, which looks at all FX trades executed by Credit Suisse over the past five years, and finds reference points broken down by currency pair, size, time of day and other factors. These are matched with a proposed trade, and used to predict the likely outcome and give a cost estimate to the client.

"There is a general trend towards transparency of FX execution within our corporate and institutional FX client base," says Cameron Mouat, London-based head of algorithmic execution for FX at Deutsche Bank, whose Autobahn platform has been offering FX algo execution since 2009. "Clients will often come to us for advice on how to conduct an upcoming trade, and an appropriate algo strategy to use – that is all part of the service. Post-trade transaction cost analysis is also a key component, giving the client feedback on the quality of execution." 

The platform uses quantitative techniques to analyse the micro-structure of the market at a given point in time, making the FX market more transparent. The quant models allow FX instruments to be analysed in a similar way to equities, from where a lot of algo execution techniques originate.

"First-generation FX algos such as TWAP and VWAP, which have more fixed-trading profiles, have been around for a few years, and we are now seeing more clients, ranging from central banks to corporates and large institutions, move towards dynamic algos based on real-time quant models," says Mr Mouat. "For example, our Stealth algo intelligently reacts to changing market conditions, liquidity and participants in order to determine how to best execute an order."

Pablo Frei, head multi-manager programmes at Pfaffikon, Switzerland-based currency manager Quaesta Capital, notes that transaction costs are a much-discussed theme in the FX market at present. Quaesta is an active FX trader itself for its currency funds, and also offers advisory services to clients.

"Many big institutions such as pension funds have been used to conducting all their trades with a single bank, but have learned that it is necessary to move to a multi-bank model in order to get the best pricing," says Mr Frei.

Quaesta offers TCA to its clients, including comparison with rates traded on the interbank market at the time of the transaction, and liquidity analysis relating to how transaction execution plays out in phases of high or low liquidity. "Depending on the size of the transaction, it may be worthwhile spreading execution out over a longer time horizon," explains Mr Frei.