ALGO TRADING IN FX.

Algorithmic trading in FX is a relatively new development but its not without its challenges. Best Execution speaks to Rob Weissman* for the inside view.

How would you characterise the FX market at present?

The FX market has been going through a unique transition in two facets: fragmentation and regulation. In the past year, the number of venues with which to trade FX and FX Options has grown dramatically, with at least five new ECNs starting up and a number of FX Option ECNs evolving. All of this growth in venues has occurred in the wake of lower overall volumes and the Dodd Frank regulation, which we expect to be enacted in the U.S. in early 2013. This will create challenges and opportunities in 2013 for most traders in finding the best destinations to trade, while adhering to regulatory requirements for FX Options, NDFs, and to a lesser extent, Forwards and Swaps.

How did algo trading develop in FX and are there parallels with its development in other asset classes such as equities?

Algo trading in FX has been growing steadily over the past few years relative to equities. Equities definitely led the way in algo trading, but because the rules of engagement are different in FX, algos have to be adapted for the specific nuances that occur in FX. An example of this is clear in trying to figure out where volume trades. As most FX destinations don’t publish trading volumes, it is not as transparent as equities. As such there is more customization required within the algos, based on the client’s liquidity sources.

What are the range and styles of algorithmic trading?

In general, the goals are to find the best liquidity sources with minimal market impact. In addition, a number of real money clients want to be able to execute against a benchmark to prove the enhancing value of their trading desk against just buying/selling at a specific time. Algos can be as simple as executing a trailing stop –with certain slippage attached – or as sophisticated as executing a multi-leg, multi-currency transaction, netted across multiple venues, based on the probability (historical standards) that you will receive better execution on one ECN versus “XYZ” bank.

How liquid an asset class is FX?

First of all, FX is a liquid asset class depending on the specific product that you are trading and when you are trading it. For instance, trying to trade one month Hungarian Forint in an Asian time zone could present some problems. When creating algos, one has to take into account the time zones they are trading, the venues in which they are trading, whether they want to be passive or aggressive and not to disrupt the individual venues, which could have a “ripple” effect on the overall market place.

What unique demands are there on algo trading technology for FX?

Algo trading technology has to take into consideration what the client wants to accomplish. Is it speed of execution, minimizing market impact, restricted broker/credit requirements, latency measurement, or multiple PB allocations across multiple funds? The algos have to be cognizant of all specific factors that the client deems important.

How would you define “Best Execution” in FX?

Best execution can be defined via TCA (Transaction Cost Analysis) tools that can analyse post trade how the client executed against the arrival price (Top of the book), arrival price (TWAP) or any other measurement based on the client’s unique liquidity pool. Due to the importance of credit, one client might have different liquidity than another client at the same time, if they have different credit profiles.

How does the FX market’s OTC nature affect algo system development?

Because of the lack of transparency in the FX market, it makes it much more difficult to build algos that are based on where volume is traded (unlike equities and futures). In addition, some liquidity pools can be more “retail-based” while others can be more “institutional-based.” This creates differences and uncorrelated flow that can be beneficial in obtaining new sources of liquidity that might not have existed a few years ago. Some platforms merge the two different liquidity pools, while others keep them separate. This creates challenges and opportunities that didn’t exist before.

Which types of clients require the greatest level of innovation to develop systems and can you give examples of their demands?

There are many types of clients that require specific demands. The most important innovation in the past year has been the push to enhance risk management as it relates to the client’s overall exposure and, if applicable, how to handle one’s clients flow.

From the sellside perspective, you want to be able to give your clients the proper tools to manage their client flows by analysing their trading behaviour. This will give them the best liquidity and allow them to internalize the flow where appropriate and have client-to-client trading, thus creating their own internal ECN.

From the buyside perspective, you want the client to be able to find the best price based on “all in” costs so you can find your best price (net of costs – ECN, Vendor, Prime Broker, etc.), and the best destinations based on probability of getting executed (reject ratios) and latency measurements. Reporting enhancements that are available now help the clients better understand where to place their trades based on size, currency pair, and time zone.