For many corporate treasury departments’ FX operations, fair and transparent execution is the overall driver. Often they use simple forwards or vanilla options and a Multi-Dealer Platform (MDP) such as 360T, Currenex or FXall. However, the growing functionality and facilities in the Single Dealer Platforms (SDP) are proving more effective for some corporate treasury departments. Yi Hahn Chin, Global Manager of Citi’s award winning e-FX platform (they just won the FX-Week award for the best e-FX Platform for corporates) CitiFX Pulse believes that, “Polling five banks at once for your FX is not the only way to achieve fair and transparent execution.” Chin has found that there are two types of clients: 1) clients who don’t monitor rates and volatility in great detail, they just want transparent and fair execution which is where the Citi Benchmark product fits; and 2) the clients who have ’sizeable’ FX transactions which if the client ask five banks on an MDP, the request may affect price and market liquidity. This is where Citi’s algorithmic trading solution fits.
Algorithmic FX order
Although, algorithmic FX trading is mainly for bigger ticket items, some corporates deploy algo trading for almost all their volume, even for 10-20 million USD trades. The allocation of flow is very much dependent on the currency pair and the risk appetite of the corporate treasury. Those who are incentivized to meet or beat a budget rate and build an audit trail around execution are the most likely adopters of the technology.
Citi assess their client’s FX business flows and risk appetite, and next develop an execution strategy for their profile, including how much of the Citi liquidity they will use in any deal. The client then submits their FX orders based on this assessment. Each FX order is a detailed description of the price limits and other aspects of what they want to achieve in the deal, through the Pulse platform, e.g. I need to buy 50 million USD selling MXN with limit price of XX and a trigger rate of 13.2250.
This transaction request is instantly passed through to an electronic blotter that is managed by the CitFX Intelligent Orders client team. They then oversee the progress of the order while it completes. Clients can watch the deal and amend/stop the deal while it is in the market. Full control over order status is crucial for ensuring that best interests and market views of the client are taken into account.
Citi produces a full report on each deal on how they filled the order at best price in the market. Each report includes:
- performance against an array of benchmarks
- time-stamped details of each child order
- costs of the service are explicitly detailed (This is essentially a full audit trail that can be stored for future reference.)
Clients are also given a graph of how the deal was made up over time showing bid rate, offer rate, limit price and executions, see example below.
Graph of algorithmic deal
Source & Copyright©2014 - Citi Corp
The deal, in the above chart, shows how the customer moved the Limit Price over the period of the execution. This deal also illustrates the completion of an order of a couple of hundred million in notional size happened without affecting the market, capturing the essence of how algorithmic trading works. It’s no wonder CitiFX call their product Intelligent Orders.
Clients mostly store the TCA (Transaction Cost Analysis) reports to demonstrate to their auditors if asked, “How did you come to this rate?”. Deal transparency is a vital feature of the service.
Citi have been providing FX benchmarks for 14 years. The principles of benchmarks haven’t changed, but the offerings have. Citi’s aim is to cater for all clients in every market.
Citi offer their core CitiFX Benchmark which is transparent set of rates audited by KPMG. Full details of the rates are published on Bloomberg and Reuters, as well as Citi’s internal sites. They also offer onshore fixing in some 20 local markets.
A recent new feature in Citi’s benchmarks is that when booking NDF crosses traders can hedge the deliverable component, against the fixing time that corresponds to the non deliverable fixing.
Today, many of Citi’s corporate clients operate in a fully automated fashion, e.g. they enter the FX needed into their TMS which is automatically sent through to the next Citi benchmark fix, take that rate and execute the deal without any other involvement. Corporates are doing this for the smaller deals and lower payment flows, e.g. from the SSC. While for the large deal sizes where hedging might be needed, then algorithmic trading is used.
Effective usage of SDPs v. MDPs
In FX, like any business dealing, there is no single solution that fits all types of FX traffic and all corporates. It’s clear that when a company needs to share their FX business between several banks, using a MDP saves time and admin. However, when the amount FX bought or sold is more than 5 million USD by value (or maybe 10 million USD) then a deal placed on a MDP asking several banks for a quote, will probably affect the market price. For really high value deals it is vital not to shift market away from you.
While FX benchmarks have their role, particularly for low value, high volume nuisance payments.
Another factor is: Citi believes that because their franchise has significant FX liquidity, that there is a less of a chance of knowing who is active in the market place when they place large deals, especially when they use algo trading.
CTMfile take: It seems counter-intuitive that a Single Dealer Platform can be more cost-effective than a Multi-Dealer Platform, because there will often? always? be the need to share FX business with several banks. However, there are hidden costs in MDPs including, the need to maintain several bank relationships and how the market price can be influenced. The quality and transparency provided in today’s SDP algorithmic FX trading and the FX benchmark systems and services raise two key questions, “Do you still need to use a MDP for FX? If so, for what.” To assume that every FX deal is done through an MDP is both unnecessary, and will probably be more expensive. This just leaves the counter-party risk aspect of SDPs.