Why you should consider increasing the algorithmic trading component of your FX execution policy

By Zeke Vince, Global Head of eFX Sales at BofA Securities, the institutional broker dealer business of Bank of America

Why you should consider increasing the algorithmic trading component of your FX execution policy
Zeke Vince

FX algo trading is on the agenda for increasing numbers of buy side firms. Why is that? 

We have found that an increasing number of buy side firms are willing to assume duration risk rather than the risk transfer price for better outcomes.  This is consistent with our trading data that shows, on average, slower execution styles can achieve more positive outcomes than the risk transfer price. 

FX algo trading can deliver rich and important datasets. Please give us some examples of these and how they can be leveraged to provide improved execution analytics? 

The most obvious example is that FX algos add FX market transparency.  Algo users have visibility into where their fills are executed and how much spread is paid.  These datasets reveal the value of real liquidity versus “recycled” liquidity, which is reflected in real-time, post-trade, and aggregate portfolio TCA.  Additionally, hourly volume profiles at the currency level may provide insight on when and how to trade, as well the cost and expected market spread. 

What types of specific trades and market conditions is algorithmic FX execution particularly suited to? 

The appeal of algos is that they provide a strategy for a range of liquidity profiles and client objectives, which allows clients to leverage a low-latency architecture to control the speed, placement and timing of execution.  There is a very strong argument that passive algos excel in volatile markets, because they primarily execute inside the bid/offer spread. On average, our data shows that slower execution styles can achieve better execution than the risk transfer price, so a passive algo strategy may be well suited to a client looking to beat risk transfer price.  On the other hand, a client who is more time sensitive may prefer a more aggressive algo that does not expose them to market movement over an extended period of time.

Many firms are starting to outsource their algo FX trading to banks. What are the benefits of doing that and what factors might influence their choice of provider? 

There are several benefits to outsourcing your FX algo trading to banks, such as gaining access to bank franchise liquidity, the ability to choose from several strategy types and the general market knowledge and experience provided by a bank’s eFX team.  Banks also have robust analytics and large data sets that provide valuable insight into the FX markets.  Additionally, banks are under intense scrutiny by regulatory bodies, so the product offerings are compliant with applicable rules and regulations. 

What advantages does algorithmic FX execution provide for clients who are particularly focused on transparency and detailed execution reporting? 

Algos offer market transparency as fills are passed directly to the client.  This is why our algos use an execution fee structure rather than a net price.  Many buy side clients need to demonstrate optimal execution and algos make it easier to do so. The datasets provided by banks to clients show that each individual fill and the bid/offer at the time of execution, given the liquidity sources available.

In what ways is the application of next generation technologies like AI and Machine Learning likely to strengthen the value proposition of FX algo trading even further? 

Advancements in AI and Machine Learning will provide clients with better market guidance, which ultimately improves pre-trade and real-time decisions during execution. Liquidity providers with access to excellent liquidity will benefit from these next-generation technologies by leveraging the quality and breadth of FX market data they collect across multiple client types and execution styles. Ultimately, advances in reinforcement learning will move us to a generation of algorithms, which are fully autonomous and are able to react to market conditions and microstructure on the fly. Users will be able to simply state an execution objective, while the algorithm determines the best way to reach that objective.