In fear of Flash Crashes: stock market wobbles past and present

Matthew Watson - Professor of Political Economy and ESRC Professorial Fellow, University of Warwick

Each event threatened the stability of the price structure of the Dow Jones Industrial Index of leading US stocks. LTCM required a private-sector bailout organised by the New York Federal Reserve Bank as its inability to exit its trading positions opened up the possibility of spreading US$1.5 trillion of uncoverable losses around US financial markets. Much, albeit by no means all, of its misfiring trading strategies centred on multiple massive bets placed on small stock price movements over a concentrated period of time. The Flash Crash, by contrast, was solely a stock market event. In one fifteen-minute period, high frequency traders managed to wipe out US$1 trillion of overall market capitalisation before backing off to allow the previous price plateau to re-emerge.

In both instances an overload of trades forced prices into a tailspin. In both instances people whose consumption possibilities are staked on the stock market through mutual fund investments stood by helplessly as their futures were determined by decisions made within the world’s biggest globally-oriented banks.

But there the similarities end.

LTCM’s problems arose from placing too much faith in the assumptions underpinning its own market models. It had constructed what, at the time, were some fairly sophisticated algorithms to determine its trading strategies. This was the dawn of the age of the quants, where the old-fashioned ‘feel’ for pricing trends was first relegated in importance below precisely calculated differences between the observed and the fully rational stock price. For as long as the good times rolled, LTCM made a fortune for its clients by throwing huge sums of money at eliminating these often minute price differentials.

Arbitrage was very much the order of the day in LTCM’s world. It assumed that it just needed to be able to stay in its positions for long enough to ensure that its trading strategies were effective. Its market models were based on algorithms that placed 100% trust in all other market participants to be as rational as it was. If its positions were profitable every time the market as a whole restored the perfectly rational price, then weren’t these, in effect, free bets? Unfortunately not, because an algorithm that can’t even recognise the possibility of alternative forms of behaviour quickly promotes activity akin to whistling in the wind when ostensibly irrational behaviour becomes the new norm. The unexpected Russian bond default in 1998 spooked all financial market participants and, amidst an ensuing loss of confidence in rational pricing dynamics, LTCM’s positions unravelled with undue haste.

Now fast forward twelve years. The Flash Crash was all about high frequency traders exploiting infinitesimal advantages in computing speed to ‘front-run’ other people’s trades. This allowed them to insert themselves between the bid and offer prices of countless stock market deals to instantaneously buy and sell the same stock in between those two buffer prices, thereby top-slicing other people’s profits. The afternoon of May 6th 2010 stood out only in the extent to which high frequency traders were successful in pushing the bid and offer prices together at the level of the latter. As a consequence, buyers disappeared temporarily from the trading environment, and prices plunged accordingly on the increasingly widespread assumption that something was up.

There was an arbitrage function of sorts going on here, but it was very different to the classic arbitrage strategy of LTCM. At most, it was so-called ‘algo-sniffing’ – trying to discover the internal characteristics of other people’s algorithms by using the advantage in computing speed to be a split second in front of them in executing their trades. None of this was about spotting a flaw in the rationality of the market as a whole and subsequently eliminating it on the demand side through weight of dollars. In an important sense – as Lewis is surely correct to point out – the high frequency traders have it within their hands to become the market, such is their capacity to hijack all potentially profitable trading positions. The strategy of choice, it seems, has shifted from trying to impose a fully rational pricing structure to trying to game all other market participants.

This is alarming.

We live in an era in which we are repeatedly asked to believe that the trajectory of stock prices tells us all we need to know about the underlying health of the economy. More and more of us have a reason to want to believe that this is true, as provision for late life is increasingly made through vicarious investments on the stock market. But, even if it was tricky to see past our own self-interest before, surely the lessons of the Flash Crash are enough to puncture any residual complacency. It proves in the most emphatic terms just how disconnected stock prices have become from underlying economic conditions.

Market fundamentalist ideology has always tried to depict the stock market as a continuous referendum on corporate performance. It is highly doubtful, however, whether any high frequency traders ever bother to waste their time learning about the companies whose stocks they are trading. The only thing that matters to them is the content of the positions that other people are attempting to take. Successful front-running requires no knowledge of the reasons for thinking that those positions make sense economically. The means through which the rest of us are now obliged to cater for our futures has increasingly been appropriated as a plaything of those whose sole concern is for magnifying existing stock price trends. We should be worried about what the future might consequently bring.