Did High Frequency Trading Contribute to the Flash Crash?
Mon, 17 May 2010 16:10:00 GMT
Much has been written in the last ten days or so about the so-called “Flash Crash”, the event that occurred in the US Equities markets on the afternoon of Thursday 6th May 2010. Many market commentators are saying that speed of the crash (and the subsequent rally) was due in large part to high frequency trading. But are they correct?
Let’s take a look at what exactly did happen during those turbulent twenty minutes of trading between 2:40pm and 3:00pm on 6th. Why did the market collapse so quickly and then recover almost as quickly? And what role did HFT and algorithmic trading actually play in all this?
These are the questions that many people have been asking. Theories abound, but it is only now that answers are starting to emerge.
Some Background
The 6th May started out like any other on the US equities markets. The Dow Jones opened down a little on the previous day’s close, then drifted lower in the first few hours of trading. Traders had an eye on Europe, as voters went to the polls for the general election in the UK and violent events were unfolding in Greece on the back of the debt crisis. Then, at around 14:35 Eastern time, things started to get a bit crazy. Within ten minutes, the Dow Jones Industrial Average (DJIA) had fallen by nearly 1,000 points. While this was happening, some stocks (e.g. Accenture, Exelon and Boston Beer Works) lost almost 100% of their value and briefly showed prices of just one penny or less, as bids completely disappeared. Then, in the space of another ten minutes, the market rallied again, to a point where it was close to its previous levels.
So what happened?
Theories That Have Been Discounted
The first theory doing the rounds was that a trader inadvertently sent an order into the market to sell a billion shares instead of a million. The rumour was that it was a fat-fingered Citigroup trader selling shares in Proctor & Gamble. Citigroup quickly released a statement saying it could find no evidence of such a trade, and so this rumour was quickly discounted.
Another theory is that the collapse was caused by some malicious hacker, but again there was no evidence to support that. Neither was there any evidence to support the idea that sophisticated cyber-terrorists were to blame.
The Likely Answer
What seems more likely (and growing evidence seems to support this theory) is that a large sell order (or more probably a series of large sell orders from a single institution) came into the S&P e-mini futures, which drove the price down and took out stop-losses along the way, causing a kind of chain-reaction as more and more stops were triggered and panic suddenly set in. Then, as traders realized there was no underlying reason for the panic, the buyback started and prices rallied. And this all happened at record speed because electronic markets are faster now than they have ever been before.
Where things went badly wrong was with those stocks that traded down to a penny. The most likely theory for why this happened is that when wave upon wave of sell orders came in, suddenly there were no more bids available on these stocks. But instead of having some mechanism to show an absence of bids, the electronic trading systems displayed bids of .01 or zero (thier default minimum value). So in this respect, was it a fundamental design fault of the exchanges’ trading systems that was to blame for much of the panic?
What About the High Frequency Traders?
One remaining question is the role the high frequency traders played in all of this. According to a number of reports, many HFT firms actually switched their systems off as prices collapsed. Manoj Narang, Founder and CEO of Tradeworx (a Hedge Fund with a high frequency trading operation), is quoted as saying he shut his HFT systems down as soon as he noticed erroneous prices, because he knew the exchange would cancel those trades (he was right, they did).
So if the majority of HFT firms pulled out of the markets while the May 6th carnage was going on, could it be argued that they had no part in contributing to the Flash Crash? Or was it their absence (and the fact that the liquidity they normally provide was suddenly withdrawn) that made things worse?
Answers on a postcard please…..

