After the Sterling flash crash on October 7th last year there were lots of theories as to why it had occurred. Maybe it was an algorithm run amok? A sort of FX version of the August 2012 Knight Capital screw up. Maybe it was a massive order executed by some large hedge fund in order to test the market and trigger stops? Or a fat finger trade? Or maybe – and with this people were really clutching at straws – it was a rational response of the market to an almost laughably anodyne statement from the French about their negotiating position on Brexit? (This was a modern day version of the claim that the 1987 stock market crash had been caused by a bad consumer confidence release.)
Whatever the reason, it was pretty clear that something major had happened. Cable (the exchange rate between British Pounds and US Dollars) really shouldn’t move 10% or so in a couple of minutes. Not least because such huge moves didn’t happen in the old, manual markets – not without substantial, meaningful releases of new information, at any rate. The BIS therefore went in to investigate, helped by the doughty Bank of England. Three months later their report is out. The conclusion? The market just kinda broke itself.
They do not appear to have found any evidence of a bug-ridden algo (despite some rumours that a large US bank’s systems had been behind much of the flow). Nor was there a particularly large order. Instead the BIS points at a number of minor causes all adding up to the breakdown.
First up, the time of day. This “played a significant role in making the sterling foreign exchange market more vulnerable to imbalances in order flow.” In Asian time (especially in the first hour or so) there are fewer traders active who have a strong view on, or trading expertise in, the GBP. Then there were option delta hedges to do and the execution of automated stop loss orders. The French announcement also might have set off a flurry of selling from algos which trade automatically off parsed headlines. All in all, pretty trivial reasons for an ‘X lifetimes of the universe’ move (under lognormal assumptions).
The fact is that the FX market, and markets in general, are getting more fragile. In my view, this is a direct result of two decades of automation – a contention that the BIS seem to agree with: “Sudden moves appear to be happening more frequently as the market becomes faster and more automated.”
Of course there is the direct impact of such automation – algorithms that execute automatically without human intervention and which are, I suspect, designed and coded very similarly in the way that they do so. After all, there aren’t so many different ways to skin that particular cat. But more interesting is the way that automation has ‘hollowed out’ human market making – the ultimate guarantor of liquidity under the old system. In part, of course, this is because algorithmic trading has taken over from humans. This is especially true in ‘fringe’ locations for particular currencies. Why have an expensive, specialist Cable trader employed in Asia (where normally no one cares) if the machine can do it?
But it is also because there are simply fewer banks willing to make markets as a result of the spiraling cost of automation creating market share concentration. Any faith that the majority of high frequency funds could add significantly to liquidity in adverse situations is misplaced in my opinion. Unlike banks, most of them they have no perceived duty to customers or the market as a whole – just to their investors. For them there are no ‘knock on’ benefits (or claimed benefits) for catching a falling knife.
It would have been nice if a particular, concrete, fixable cause could have been found for this event. The fact that it wasn’t confirms that it was a case of endogenous fragility. This is something that I am not wholly surprised at – as anyone who has heard or read me rant on about this subject could no doubt guess.
So what conclusions do the BIS draw as to what to do? For starters, regulators need to do some homework: “it is important for policymakers to continue to develop a deeper understanding of modern market structure and its associated vulnerabilities”, according to Guy DeBelle (who is an extremely sensible cove in my estimation). And how about market participants? They should “consider how to ride these [moves] out” says the report. Well, that’s comforting.
In conclusion, after three months of work from the BIS we can summarise the findings as follows: shit happens, so get used to it and put on your seat belts.