Big brokers began offering foreign-exchange trading algorithms five years ago, and now all the major players have an offering. Now a vendor is hoping to upend the status quo with a suite of algorithms it claims will offer better execution.
“We are an independent provider of trading technology, and there’s a greater awareness these days in the FX market about existing algo providers who are generally dealer banks with a principal interest,” said David Mechner, chief executive at Pragma, historically a vendor of equities algorithms. “Their P&L is in direct conflict with that of the client.”
Mechner argues that banks may favor their own liquidity pools when handling customer orders and that may work to the disadvantage of the customer. “It’s an obvious conflict of interest,” he said. “A lot of the algo offerings the banks provide explicitly trade into their own stream. Some of them are mixed where they may or may not. Some will mainly trade on ECNs, but there’s a clear awareness that there’s a role that an independent firm can fill there.”
[See Traders Magazine Forex Coverage]
Foreign-exchange algorithms, like their cousins in the equity market, break up a large block trade and feed it piecemeal into the marketplace over time. In the FX world, the algos succeed point-and-click aggregation technology that simply gives the buyside access to a big pool of liquidity that might include ECNs as well as dealers. They are point-and-click mechanisms much like the old direct-market-access platforms of the traditional equities world.
Pragma is offering algorithms and not aggregation technology. It is pursuing hedge funds and commodity trading advisors as customers. Mechner contends this group trades more FX than traditional buysiders, who largely leave the trading to their custodians. A formal product launch is slated for the first quarter of next year.
GREATER ALGO EFFICIENCY
An early customer is Baltimore-based Campbell & Co., a systematic investment manager offering a wide range of strategies to clients. Roughly one-third of the firm’s $3.2 billion in assets under management get invested in G10 and some emerging markets currencies. About 90 percent or more of the firm’s FX trading is done electronically, said John Radle, Campbell’s global head of trading.
The firm was one of the early adopters of broker-supplied FX algos, and uses them to trade large orders to reduce market impact and because it makes the trading desk more efficient, Radle said.
Campbell’s pursuit of best execution led it to tap Pragma to develop an FX algo solution. Pragma is creating a liquidity pool comprised of connections to about 10 banks and five ECNs that Campbell has selected. With its expertise in developing algos, the firm is also creating algos for Campbell designed to achieve the CTA’s benchmarks, Radle said.
For instance, if Campbell wants to do a short-term time-weighted average price trade, a bank algo may trade an order in slices of $1 million, he said. Pragma may break the order into smaller slices to send to ECNs to provide a better approximation of TWAP over the period, he said.
“We feel the Pragma relationship can allow us to get a number of best-of-breed bank and ECN feeds and hopefully execute a superior strategy in a superior liquidity pool, and ultimately that will get us better results,” Radle said.
“We will be sending parent-level orders to Pragma, and they will do the slicing and dicing within that pool of ECN and bank liquidity that we custom-created,” he said.
The firm is hoping to get better executions to help lower transaction costs with the new solution, Radle said, adding that Campbell will compare the Pragma and bank algos to see which performs best.
In addition, Campbell will use Pragma’s new FX TCA offering to measure the algos’ performance and help ensure the trading costs it gets when the strategies go into production are at least as good as the estimated costs it saw when the strategies were being back-tested.
Pragma executives say banks offering algos have a different economic interest from just providing best execution, because they make money from proprietary trading and internalization. Orders never leave their premises.
That resonates with one buyside trader who is concerned about loss of control. Michael O’Brien, head of global trading at Eaton Vance, said his concern about bank algos is that when he clicks the submit button to send an algo, he’s losing control over where the order goes. “Is it going to ECNs, how quickly, is it just waiting for an internal match?” he said. “Some banks will explicitly tell you they’ll do both. It’s impossible for me to verify.”
In response, Citi’s James Dalton said that post-trade analysis is “completely transparent and provides clients the ability to gauge the value of using algorithms over time.” Dalton is a director for Citi who built and runs the bank’s CitiFX Intelligent Orders, the firm’s FX algo platform.
Citi first rolled out algos in 2007. The first wave of product was designed to help traders shift larger orders through the market without disrupting it, he said. These algos helped clients control how much liquidity they took from the market while executing to ensure they were never taking so much that they were driving the market, he said. Over time, that’s evolved so the algos can adjust in real time according to what is going on in the market, Dalton said.
BEYOND THE NICHE
Meanwhile, as the range of clients expanded from institutions to include corporate clients, sovereigns and other banks, Citi needed to provide for a greater range of execution styles. The firm still offers passive strategies, including Silent Partner, which executes in the interdealer bank market, and Ripple, which executes in Citi’s own liquidity pool, but has also added aggressive strategies used more by spot traders or people who are actively trading intraday, Dalton said.
FX algos are emerging from a position as a niche service as the breadth of customers using the tools has grown, Dalton said. Growth has been fueled as an equity execution style has filtered across the FX market, and as firms look for the audit trail that an algo can provide. “It’s still not used universally because some firms don’t trade in an order size that would make the cost benefit worthwhile,” and others haven’t set up their systems processes to be able to automate workflow around the use of algorithms, Dalton said.
Jonathan Wykes, head of eFX sales EMEA at Bank of America Merrill Lynch in London, echoes Dalton. He said clients can receive a post-trade report on execution outcomes. The TCA “details every fill they got, where it was executed, the bid, mid and offer,” he noted.
Merrill recently rolled out a suite of algorithms for clients under its FX Instinct brand. These algos are already used by the firm’s trading desks in G10 and EM FX trading to access both external liquidity and the bank’s own liquidity pools. The suite includes both passive and aggressive strategies that can now be used by corporates and banks. The algos are already available via Bloomberg and will soon be available on FXall, FX Connect and Portware.
Merrill also offers what it calls a “one-touch consultative service” where the desk enters an algo order on the client’s behalf. Clients use this when they aren’t sure which strategy to use and advice is helpful, Wykes said.
Like Merrill, Barclays will soon be rolling out algos for clients. The broker operates an aggregation platform called Gator that provides access to the key FX external market liquidity sources as well as to the firm’s internal pool. It was originally built for the firm’s own FX trading desk and includes a display of liquidity across markets that the firm believes is a “true representation of the underlying liquidity in the market,” according to Holden Sibley, head of North American BARX sales for Barclays.
The next step will be to build out an array of algos to access Gator liquidity. Currently, the firm is offering basic order types to tap the Gator liquidity, Sibley said. Barclays already offers algos, rolled out in 2006, to access the firm’s internal liquidity. Both the algos and Gator are part of the firm’s BARX FX electronic platform.
