The Last Look…
Posted by Colin Lambert. Last updated: February 13, 2023
Long-term readers will be happy to hear that after a prolonged spell away, it’s time for me to take on the world of academia again!
Although it’s been a while since some of my friends in the prop trading space bothered me, a couple were in touch last week after a Bloomberg News report cited a paper by UK regulator the Financial Conduct Authority that basically says HFTs provide more order book depth than banks.
I must confess, I read the report when it came out a few weeks ago, and didn’t give it much thought, but as people are trying to leverage it into a statement of support for the HFT model, I feel obliged to comment!
Before loading up, I should stress that the report uses a very important phrase, “in normal times”. This can, of course, mean anything, but generally the report is talking liquidity shocks for abnormal conditions. Equally, it does highlight the different nature of liquidity provision by HFTs and banks – I guess my thought at the time was “who cares about HFT liquidity?”
I find the report yet another example of worthy work by academics who are either too tied into the equity markets model and their obsession with retail investors, or who don’t understand the basic concepts of best execution as it pertains to FX. I don’t think there are any doubts that these firms offer more order book liquidity at top-of-book, especially on the “primaries”, but look at what has happened to those venues over the years as these firms have taken over. Volume has dropped off dramatically. In 2011, EBS was executing $159 billion per day – and this is prior to EBS Direct and the other new venues – and then-Thomson Reuters was trading $150 billion per day, before it bought FXall.
Those are stark declines, and while it could be argued that the HFTs have stepped into a gap left by the banks, there is little doubt, anecdotally I have to accept, that a big reason for the decline in activity was the rise of signalling risk (largely seen as giving information to smaller, nimbler HFTs on these venues) and a perceived decline in the general trading environment. Both venues have made efforts in recent years to improve that environment, which just reinforces the point that many (but certainly not all) of the HFTs have made it a worse place to trade.
It is this that I think the academics miss. It’s all very well being a bigger part of a smaller pie, but what about the damage that some of these firms have done to create those conditions? Larger tickets are rarely going through these venues because of the “games” played by other LPs.
It could be argued that firms like XTX Markets and Citadel Securities have improved the CLOBs, and I would be OK with that, although I don’t think they can be referred to as “HFTs” in the manner the paper describes them. Maybe, just maybe, what the researchers have picked up upon is more volume from these firms and mis-labelled them?
I cannot stress enough that FX and equities are different beasts – this makes the best execution requirements vastly different
It is also relevant, in my view, that firm liquidity in the public spot FX market is probably somewhere in the region of $80-90 billion per day (for reference I am using $25-30 billion on the primaries, $20 billion on LMAX Exchange and $10-15 billion on Cboe FX firm). That’s around, probably under, 4% of total spot volume as per the BIS.
The top-level numbers, therefore, are not that impressive, but what about the more important execution quality? This is another example of someone thinking that either everybody wants to deal in one million units (or less), or that you can do as much as you like at top-of-book, because the FX market is so liquid, there is great depth of book. Both are fallacies, although I accept that a lot of hedging and execution is done in smaller amounts these days, that is more a factor of the market structure change.
I cannot stress enough that FX and equities are different beasts – the latter is largely a retail business, albeit aggregated to a degree, while the latter remains an institutional one (just $194 billion per day is traded by retail clients across all products). This makes the best execution requirements vastly different.
A retail client wants to hit top-of-book in their amount and can often do so. In FX, a client may have hundreds of millions to do – the last place they want to hit using an algo or otherwise, is a public market in one million. Look up signalling risk and market impact in a financial dictionary, this will probably be the definition.
The fact remains, if we look at the core reason for the FX market – “real economy” end users – they don’t want to execute publicly due to market impact, and they certainly don’t want to face an HFT who, by its very nature, will be hedging out of the flow within milliseconds and exacerbated the impact of a relatively small trade.
The paper, ultimately, is just another academic exercise that in my mind ignores the reality of the FX market. Yes, without doubt, as the paper observes, HFTs help iron out “price inefficiencies”, but does that really help the majority of players? They do so using speed alone, not by taking on actual market risk.
An insight into the thinking can be found in the following line from the conclusion of the paper, which incidentally can be read here, “…we find HFTs’ order-book liquidity provision is less sensitive to market-wide volatility spikes. In contrast, dealers’ order-book liquidity pro- vision is more resilient ahead of discrete single-security volatility, in the form of scheduled macroeconomic news announcements.”
Since when did FX become a “single-security”?