By Jay Hurley
Jay Hurley, FPL Global Foreign Exchange Committee Co-Chair and Vice President, Morgan Stanley Fixed Income, unravels FIX’s role in FX and argues for FX to be integrated into equities algo trading.
Adoption of FIX in FX
Adoption of FIX in FX for the core functions, such as streaming prices, orders and executions, has done well. Evolving from a situation where few ECN’s and banks used FIX for FX to one where the last ECN not using FIX will be FIX compatible in January of 2011. While adoption at the ECN level is almost 100%, for banks it is probably around 70%, up from 15% 5 years ago.
I would say FIX sold itself. It just naturally happens that when you get enough critical mass, after that tipping point, outliers start to look unusual. From there, it becomes more important for FIX to address more than just core functions. For instance, nondeliverable forwards (NDFs), which include most of the Asian currencies that cannot be freely traded, have some of their own specific features that need to be incorporated into the protocol.
New Developments in FIX for FX
The FIX Protocol for FX has provided an opportunity for rapid product development and deployment, and in doing so has increased competition in the market space. FIX provides the flexibility necessary for a platform provider to work with potential users to provide product enhancements. If a product doesn’t meet the needs of the market participants, it will fail quickly; but other times a new product will fill a gap and become the new force in FX trading. This process does not need regulatory oversight for it to happen – it happens by innovation.
Currently, the GFXC is working on OTC options as well. FX specific issues for options revolve around the fact that often at the end of the day, traders receive deliverable cash, so questions like ‘What is the currency of your option?’ are not as obvious. In a currency option, you have several currencies: the currency of the option, the currency that it is against and the currency of the payment, which can be a third currency.
The FPL Global Foreign Exchange Committee (GFXC) has also worked on FIX for allocations, which has some quirks for FX that are not present in equities. For example, if a fund manager has 50 sub-funds, often they will do a single FX trade, representing the net exposure of all of the funds. Rather than buying and selling some all day, they aggregate it. As a result, a single trade of ‘buy 1 million Euros, sell US Dollars,’ turns into an allocation of sell 50 million Euros, buy 150, sell 100, etc. Also, the net trade cannot be zero; otherwise, you cannot settle the trade. This issue caused a lot of consternation, but because there is still a need for a way to do the allocations in a trade where the net is very small, it was decided to have a one cent minimum.
Other factors FIX has to address include delivery and settlement dates, and NDFs cannot actually be delivered, so they are cash settled – normally in US Dollars. There are also questions regarding what fixing rate to use, because FX is an OTC market and there can be several semi-official prices to choose from.
HFT and Algorithmic trading in FX
The major current trend in FX is similar to the evolution of equities, which is towards more High Frequency Trading. Essentially, traditional equities brokers have access to all the same ECNs and exchanges as each other and the prices are all offical, so added value comes from how smart a broker’s algo is and how much potential liquidity they have in their internal pool. Whereas, in FX internal liquidity matters more than which ECN’s you have access to, because a trader can make a price from inventory without access to an ECN.
Algorithms in FX operate on a different basis than equities, where an algo winds its way through smart-order routing and dark pools to get the best execution. The real goal for bank FX liquidity providers is, for them to build their own internal liquidity pool and have 100% of their trades match within their group. Essentially, they could find buyers and sellers amongst their different clients. Some of them would run algos, while some would ask for one-off prices and it would all just transact. It is not ever going to reach that extent, but a lot of the top players have so much going through their own books that their prices are heavily skewed towards their client flow.
Running an algo with one broker could get a totally different price than running it with another because the prices one trades against would be different that the other’s. Then, because there is no official price and there are no volume figures, how do you do a VWAP trade?
It is difficult for FX traders to trust algos. Just as in equities, half the work is post-trade analysis. By the time a trader runs an algo, it will be optimized against historical data, so they will run an impact analysis to measure today’s performance. There might be too much impact, however, simply because they traded with a bank who happened to be long and they made the counterparty more long, but if they traded with someone who was short, there would be much less impact. It is difficult for people to trust something that they cannot definitively say is working. There are obvious algo applications that can reduce market impact and execution cost, but once you get past the obvious, if you cannot measure it, it does not really exist.
FIX and the FX Back Office
FIX is unlikely to have a huge impact on the back office; it is possible, but unlikely. Although FIX adoption has been good on the core trading platform, I think allocations would take some time to be supported because many traders are not using an OMS/ EMS. It is remarkably common for FX traders to use their platform on their desk all day and then fire off an Excel spreadsheet or an email to allocate their trade at the end of day. We have the allocations FIX spec to enable more automation of this work flow.
FX had a good financial crisis. Even though the market was volatile in the extreme and there were various lockups in sovereign bonds and CDS’s, it functioned perfectly fine despite everything that happened. On May 6, for example was very busy: I did what was typically half our entire 24-hour volume in the Asia morning. While it was busy, it all continued to function.
I think FIX is a solution to the FX problem that people do not know that they have. There is still a lack of appreciation for the effect that FX trading can have on the return of your equity portfolio. Let us take the Australian Dollar (AUD) as an example, which in 2010, has gone up 50% from its lows in October 2008. If a trader made a call on Australian stocks going down, and they were short, a downward movement in the stock price of 10% with an increase of 50% in the currency would have resulted in a greater loss due to the currency swing than would have been made on the equity portion of the trade. Traders may look at a TWAP and analyse the execution cost as a few basis points, but if the currency moved against them 3%, does even ten basis points really matter more than the 3%?
Traders need to look at their work in a currency-hedged fashion. Eventually, it has to reach a point where equity and FX trading are synchronized. At that point you could run a FIX algorithmic trading definition language (FIXatdl) order and attach an FX instruction on the side of it. This is technically possible, but the problem is no one would use it because they do not know that they ought to. There is no point in putting the cart miles down the road in front of the horse, just yet, but there is huge potential for more cross asset class trading.
Thomson Reuters’ Ned Micelli shares the benefits of including FIX in FX matching services.
The Financial Information eXchange (FIX) Protocol provides an industry proven, highly efficient mechanism for order management and post-trade workflows. While FIX has traditionally been the protocol of choice for equities products, there has been an increasing demand from market participants to utilize FIX for FX trading workflows. We see the usage of FIX as being a key enabler for algorithmic FX trading into our matching service. The use of a low latency FX-FIX interface allows customers to remove legacy hardware from their site, while benefiting from an industry standard FIX interface. Removing our client’s dependency on Thomson Reuters’ provided hardware also allows our FX customers to quickly embrace turn key hosting solutions. This approach allows FX traders to readily adapt their existing FIX based algorithmic trading engines into our FX matching service while, at the same time, reducing their hardware footprint and their overall cost of ownership. Our matching service will also use FIX for post-trade processing, where standard Trade Capture Reports (TCRs) will be used to represent each FX trade executed.