Navinder Singh Sarao lost his attempt to resist extradition from the UK to the USA in mid-October. He was the man trading from the bedroom of his parents’ house in some north-west London suburb whom the US regulators believe was responsible for the Wall Street flash crash of May 2010. If their belief in his guilt is upheld by the courts, he faces a prison sentence of up to 350 years. That’s substantially longer than the 150 years scored by Bernie Madoff, by the way. Incidentally, I’ve often wondered about these long sentences handed down by American judges – either they are a polite way of saying “we’re throwing away the key”, or the US legal profession has a touching faith in the ability of scientists imminently to prolong the normal human lifespan. Or maybe they think the atmosphere and diet of American gaols is so healthy…no, let’s not go down that road.
‘Painting the Picture’
Anyway, that’s what Sarao faces; now, the case hasn’t been heard, and all we have are various reports of what may or may not have taken place, but it’s still interesting to look a little at the proposition that he was solely responsible for that 1000 point dip in the Dow. What has been suggested is that he was running a spoofing programme, where he would continually load sell orders fractions above the ruling market in order to give the impression of a greater volume of selling than was actually the case. He would lift those orders if buying approached too close and it looked as though they may be triggered. In other words, in the old market phrase, he was “painting the picture”. The end-game – where the profits should have come – was that once his implied selling had persuaded others to put in real selling and the price had dropped, he could buy the market, withdraw his fictional sell orders and watch the ensuing short-covering bring him his reward. Well, that wouldn’t be that unusual; it’s a strategy that’s been used again and again over the last hundred and whatever years. Sometimes it would succeed, sometimes not, depending really upon the accuracy of the trader’s feel for how much real buying and selling there was.
But this time, there seems to have been something else at play as well. After all, the drop in the index was savage, far greater than would normally be expected from a (I guess) pretty small day-trader playing games. The first thing to look at is that it is suggested that in the period leading up to May 2010, Sarao had been using an automated algorithmic programme to manage the spoof orders he was putting in; the way that could influence market movement – although I would certainly go no stronger than could – would be on the straightforward basis that the algo would react quicker than a human trader, thus potentially giving a greater edge to the spoof.
But the more interesting consideration is one that comes out of a report recently produced by the Bundesbank. They have been looking at algo trading and HFTs and they have come to the conclusion that there are fundamentally two categories; those that trade actively on news, and those that act (‘passively’) as market-makers between buyers and sellers of assets. The findings were that the first category are active during periods of market volatility, and in themselves contribute to that volatility. The second category, often providing market liquidity during quiet periods, tend to withdraw from the market when volatility increases, which is precisely the time when that liquidity would be most desired. So on the one hand, those following news, and therefore likely to be trading in a similar direction, will be there and at just that time, those who would provide liquidity to absorb the first categories’ trading are standing back from the market. Well, that sounds pretty reasonable, and not terribly different from what we may have expected from non-automated markets in the past. Where there is a difference, though, is in the subsequent behaviour. Trend following algos do precisely that – they follow trends; so the movement – as in May 2010 – risks being dramatically over-extended, because while the trend is running up or down, the machines will keep following it, adding substantially to volatility.
Elephant in the Room
That may be good or may be bad, depending upon where you stand, but I think the regulators have a problem here. Essentially, they mostly like automated trading; it provides a clear audit trail and that gives a greater confidence in being able to follow what happened. But the elephant sitting in the regulators’ room is the way that that very same trading that gives clarity also works – at the most critical time – to reduce liquidity. This problem is only going to grow; the influence of computing power and particularly AI is going to dominate financial market trading in the rapidly approaching future. Most of the time, that works for us all – let’s face it, if you can get a machine to do something, why not? But the risk is the occasional potentially catastrophic move where the machines are uncontrollable. Nobody’s really got hold of this yet. Yes, you can put in circuit-breakers and trading pauses, you can impose time limits on unfilled orders, and yes, MiFID 11 may help, but problems will recur as long as operational power runs ahead of regulatory understanding.
This month is the thirtieth anniversary of the ‘Big Bang’ in the City of London; for me – and I would hazard a guess for most of the readers of this column – it’s been a huge benefit in terms of our careers. However, one of the central issues it raised, which has still not been adequately resolved, is what happens when you replace personal responsibility with legal instruction and change the fundamental question in the trader’s mind from ‘is it right or wrong?’ to ‘is it legal or illegal?’ The machines can’t answer the first one, so we have to be ahead of them in defining the second – and we’re probably not.
And Navinder Singh Sarao? Well, he seems like a small fish (although I stress the the case has still not been heard) and there’s a populist demand for the blood of bankers and financial market traders, so I have a nasty feeling he may be hung out to dry. Otherwise, the regulatory authority would have to admit that it was running behind the practitioners.
But 350 years? I’ll never understand the USA.