In London, burger joints have proliferated hugely in recent years. As result, there are simply too many and unfortunately some have closed (including my local favourite, a branch of Gourmet Burger Kitchen, this week). Despite, their proliferation, it seems as though we simply aren’t eating enough burgers, despite the fact, that I’m personally trying to do my bit, burger-wise.
In the financial markets, it can sometimes seem that way, where some theme seems to be everywhere. Yet, in practice it hasn’t really been embraced by market participants. When it comes to ideas such as alternative data, is this the case (and I’m currently co-authoring a book about the topic with Alexander Denev)?. Whilst there is a lot of chatter about alternative data, it’s only really the major quant funds who have been using it for years, whilst other firms are playing catch up. A lot of the growth in alternative data is likely ahead of us.
Is the same true true of transaction cost analysis in recent years? TCA basically involves analysis which can work out how much you’ve paid for a trade. Recently, there has been an uptake in TCA helped by regulation, such as MiFID II, which discusses best execution. In other words, the buy side needs to show that they have got the best price (at least in some asset classes). Is everyone doing TCA yet or is more a case of cheeseburgers without cheese patties..? Greenwich Associates recently released their 2019 report on transaction cost analysis to help answer this question (summary here and also here) – thanks Wilfred Daye for posting their press release on LinkedIn! The report which surveyed 256 firms in Europe and North America, noted that 95% of trading desks in their sample are using TCA for equities. The high take up for equities perhaps isn’t such a big surprise. However, in other asset classes, usage is comparatively low, indeed in FX it’s 63% and much lower at 37% in fixed income. Indeed, this is probably reflected in the number of TCA solutions there are in equities versus other asset classes. If you’re interested in a TCA solution, tcapy, Cuemacro’s TCA software is now available and is focused on FX spot (more details here) and it can be expanded to other asset classes too.
There are of course complexities in FX, compared to equities. The FX market is more fragmented than equities, and there isn’t a consolidated trade tape (and I doubt there will be a fully comprehensive one any time soon), although there are some some interesting new products such as FastMatch’s TradeTape, which do try to address this. Whilst FX is quite fragmented it is also now possible to get comprehensive hourly volume data from CLS (and also some FX flow data), so you can get an idea of how large your flow is versus the volume in that currency pair. For deliverable currencies which are settled by CLS, their coverage is over 50% of the market for volume.
Typically, FX is pretty liquid, especially in G10 FX, so there isn’t the issue that there aren’t any quotes to construct a benchmark. In practice, we might wish to construct a benchmark using a number of different price streams to take into account our various liquidity providers and the venues FX trades on.
In fixed income, there are also issues, which can make TCA more challenging. Many of the assets within fixed income are less liquid and it can be difficult to know what a market benchmark price is (eg. off-the-run bonds). However, you do have TRACE data in the USA which records fixed income transaction data.
The argument that, we shouldn’t do TCA in FX or fixed income, because it’s a bit more difficult, isn’t one that I agree with. Yes, having high frequency data to benchmark against is the ideal situation, which might not always be available for all fixed income instruments, but even having the open, high, low and close quotes, will give us some way of comparison. For example, if we repeatedly get filled for sells at the low of the day, that is a valuable observation if you’re trading.
Ultimately, TCA shouldn’t just be about box ticking for the assets where it’s mandated, it should go beyond that and encompass as many of the assets that you can trade, even if TCA isn’t “easy” in those assets. If you save money on how you execute, that is good for you and for your investors. That’s why it’s worth spending time on TCA and using a solution that you can customise to your requirements, like Cuemacro’s tcapy Python based library. Using tcapy means you don’t have to reinvent the wheel and also allows you to keep your trade data private.