By Anne Friberg
Corporate FX traders can benefit from “algos” to improve execution performance and reduce transaction costs. International Treasurer sits down for a Q&A with Citi.
Algorithmic trading may seem like a strategy best kept among the “quants” in the hedge-fund world, but as this recent interview with Citi shows, corporate traders can benefit in very measurable ways from algorithms offered by their FX banks.
What are the main benefits companies can derive from algorithmic trading? Algos provide more efficient execution for customers by minimizing transaction costs in two ways, explained Andrew Tokar, director of corporate FX eSolutions at Citi.
First, he said, algos reduce or eliminate information leakage that can result from poor execution, and second, a trade can be divided in multiple pieces that can be executed over a period of time at the lowest possible spreads available in the market, across both public and private venues (the latter of which companies can’t usually access).
Companies that face large, episodic risks can benefit greatly from algorithmic execution
Another benefit is a higher level of transparency, Mr. Tokar said. That’s because for any algo strategy, the bank provides the full detail of each trade, time-stamping each individual “clip,” and showing the all-in rate achieved benchmarked against the [primary] market inception price, as well as the market average price over the duration of the execution period.
Access, Systems and STP
Clients can access the CitiFX Intelligence Orders (IO) suite using any of Citi’s proprietary platforms (CitiFX Pulse, CitiFX Velocity, and via CitiFX API), as well as on select third-party platforms, e.g., Bloomberg.
To the extent a customer’s ERP or TMS is already integrated with any of those to achieve straight-through processing (STP), using an algo vs. any other trade execution form would not add any process step or re-keying of trade details.
For those who have the STP setup via FXall or 360T, it would require re-keying into one of the interfaces above. For the purposes of confirmation and settlement, all the pieces of the algo trade get aggregated into one trade.
What type of company benefits the most from algos? Companies that face large, episodic risks, say, settling an M&A deal in foreign currency, can benefit greatly from algorithmic execution (large can also mean relative to the daily liquidity of the currency pair in question). Although the FX market trades about $4 trillion per day, large executions needs to be handled carefully. Stealthy execution is crucial in a market that is only too ready to take advantage of a price-taking hedger whose objectives are known. Companies who regularly enter the market can also benefit from algorithmic execution on their portfolio of day-to-day trades, almost no matter the trade size. Over time, there can be meaningful cost savings from algos on a representative “auto-pilot” portfolio of trades compared to typical price-taking on a multibank ECN.
How do algos work? Practically speaking, large orders are broken up into smaller increments, each of which is managed by an intelligent process prescribed by the selected algorithm. Note that this means that the risk transfer from client to bank also occurs incrementally. Citi has branded its suite of algos CitiFX Intelligent Orders, IO, and it includes active/aggressive and passive, “agency” and “principal” strategies. Between agency and principal algorithms, Citi does not favor one at the expense of the other. Clients can take liquidity either from the wider market via direct market access (DMA) or from Citi’s own liquidity pool (internalization), depending on their objectives. In any case, the bank and client agree upfront on a spread to be applied to intelligent orders.
The two most popular aglo strategies are Silent Partner and Ripple:
Passive strategies typically lend themselves better for use by corporates looking for efficient hedge execution as they are designed to calculate the highest percentage of market liquidity to use to ensure the fastest execution possible without alerting the market. The two most popular are Silent Partner (DMA model) and Ripple (internalization). For other strategies, see sidebar below:
More Algo Types
Trading algorithms are often given descriptive names inspired by what they are designed to achieve. Here are a couple more examples of FX algos offered by Citi:
Dagger: An aggressive, direct market access (DMA) strategy designed to consume liquidity from all sources, at a rate equal to or better than a client-defined limit price.
Liquid Slice: A passive, direct market access (DMA), time-weighted average price (TWAP) strategy designed to get as much of the trade filled on the passive side of the market, helping achieve a smooth average for a target amount over a client-designated time horizon.
Silent Partner is an agency model that employs smart-order routing to directly access the interbank market (normally not accessible to corporate customers) and other liquidity pools like EBS, Currenex, CME and Reuters.
The algorithm works passive orders in the direct market, with the size, frequency, and timing of each execution based on granular historical data which indicates how much liquidity can be consumed at any time without the market moving against the client. If fill ratios begin to lag, then Silent Partner briefly shifts to a more aggressive mode to keep pace. The aim is to always trade at the lowest possible cost for the client; this means that sometimes the cheapest liquidity might be found in the bank’s own flows and sometimes in the interbank market.
Ripple is an internalization strategy that offsets customer orders with the complementary flows Citi sees via its electronic price-distribution network. Ripple looks to work orders within the interbank bid/offer spread for each currency pair at a rate of execution that takes prevailing liquidity conditions into account but without going to the external market.
Ripple ensures the speed and method of execution minimize the impact (or ripple effect) of each order on the market. The bank internalizes as much of the risk as possible by fragmenting positions between a portfolio of related crosses and correlated pairs. This allows Citi to reduce its own risk quietly and more efficiently, too.
The table and chart below are components of the post-trade reporting to a customer who used Citi’s Ripple algorithm to sell an Aussie dollar position.
(Editor’s note: A portion of this material appeared in Citi’s CSG Monthly. Citi will sponsor The NeuGroup’s FX Managers’ Peer Group 2 meeting in September)