Yesterday, it was reported in the Financial Times that several FX traders who were charged with manipulating the WMR fixing at 4pm LDN were found not guilty in a New York court. Offering a WMR fixing service actually results in a market maker taking substantial risk on a transaction, which seems to have been lost in a lot of the commentary about WMR. Below, I’m going to give a quick summary of what the WMR fixing is and will also explain why the market maker could be taking risks on that. Later, I’ll explain the benefits of using TCA in general to understand your execution costs.
Unlike other markets, there is no traditional daily market close in the FX markets. For G10, the markets open on Sunday in Auckland time and keep trading till New York close on the following Friday. The WMR fixing historically came about because of client demand to be able to execute at something like a “close”. A client can place a fixing order (say at 3:30pm LDN) and the market maker will guarantee to give the client whatever the fixing price ends up as at 4pm LDN. Of course, a market maker is taking a risk on this transaction, because they don’t know what the price will be. Essentially the market maker is basically selling an intraday option to the client. I wrote a paper on this subject a few years ago, which was cited in the Wall Street Journal (link here). This cannot be a “free” option. Even if there are cases where trading at the fix may be profitable for a market maker, there will be instances where the marker maker could be totally steamrollered by flow against them. There has to be some way of compensating the market maker for taking this unknown risk (a risk premium), otherwise they will no longer be able to offer this service. If the market maker keeps losing money, they will go out of business, which isn’t going to be helpful in the long term for the prospects of offering market liquidity to their clients.
Is the WMR fixing optimal for every execution by a client? That’s an open question, although I suspect not. The reason to use a fixing order for a client is because it’s simpler and in some cases a client might be benchmarked against the WMR fixing. It is not necessarily that it is “cheapest” in terms of cost. If you execute at the fix, you are executing over a relatively small window of the day. Sure, execute over a small window if you have specific view that the market will start to move against you soon. However, if you don’t have a specific view about the direction, then it probably makes sense to execute over a longer window, where potentially your flow is a smaller amount of the overall volume?
What the whole WMR subject does bring to mind is that if you want to get better execution you need to spend time analysing what you are doing in your executions. This is not possible to do just looking at one trade. It needs to be done over a long period of time, to get a good sample of your trades. Is executing at the fix better for you or at other times of the day? When is liquidity best? We can use transaction cost analysis to find that out over a large history of your trades, understanding metrics such as slippage (difference between your executed price and the market mid) and market impact etc. Are some brokers better at executing certain currency pairs? Again, we can use transaction cost analysis to do that.
Over the past year, I’ve been working on a Python based FX TCA library. I’m hoping to open source the basic framework of it, if I can get sponsorship (sponsors will get access to additional features and support). I’m hoping by making TCA more accessible and transparent, by making the source code available. It will empower traders to be able to do their own customised TCA, tailored to their trade flow. If you are interested in supporting my TCA project, and want to make TCA easier and more accessible to folks, let me know!