Could high-frequency traders cause another flash crash?

Market prices increasingly driven by computer algorithms, research indicates

The growth of computerised trading has brought obvious benefits to investors. Trading costs have been slashed; complex information is quickly processed and incorporated into financial prices; and markets are more liquid and more accessible than ever before.

It’s not all sunshine, however. There are concerns markets are increasingly being moved by computer algorithms rather than by any fundamental catalyst, resulting in increased volatility and potentially causing instability or even crashes.

Rise of the algorithms

Fears have largely centred on high- frequency trading (HFT) firms, which use complex computer algorithms to execute trades at lightning speeds. The rise of HFT – it now accounts for the bulk of daily trading activity in US and European markets – has resulted in average holding periods for stocks plummeting to an estimated 22 seconds.

HFT has been in the spotlight ever since last year's publication of Michael Lewis's Flash Boys, which took a dim view of the practice. Shortly afterwards, the Federal Bureau of Investigation announced it was investigating HFT firms.

In April, London trader Navinder Sarao was arrested and accused of having contributed to the infamous "flash crash" of May 6th, 2010, when almost $1 trillion in market value was lost in a matter of minutes. In the same month, the Senior Supervisors Group, a group of financial regulators from 10 countries, warned HFT increases the potential for "systemic risk" to develop "over very short periods of time". In May, the Bank of England's Prudential Regulation Authority announced algorithmic trading firms would be scrutinised to ensure their controls were "fit for purpose".

Bogus orders Despite regulators’ tough words, the suspicion remains they are not on top of the issue, a suspicion that is not allayed by the arrest of

Sarao. Authorities accuse Sarao of spoofing – flooding the market with bogus orders in order to mislead other traders. On the day of the flash crash, authorities say, Sarao was responsible for one in every five sell orders, introducing “artificial volatility” into the market.

Sarao’s arrest creates more questions than it answers. Regulators’ September 2010 report into the flash crash makes no reference to Sarao or to spoofing; then, authorities pinned the blame on a faulty algorithm that accidentally unleashed a flood of selling.

Why did it take so long for the arrest? Why have no other parties been charged, given that spoofing is a notoriously common practice among HFT firms (Sarao himself complained to the Chicago Mercantile Exchange on more than 100 occasions regarding HFT practices he regarded as manipulative)? What was so different about the day of the flash crash to other days? If it is true that a self-employed stay- at-home trader could unleash market mayhem, what does that say about the fragility of the financial system?


While we may never know the exact trigger for the events of May 2010, researchers fear similar episodes are likely. A 2011 UK government report, for example, cautioned computer-based trading may lead to a different financial system “in which crises and critical events are more likely to occur in the first place, even in the absence of larger or more frequent external fundamental shocks”. Trading volumes and prices may become “prone to cascades, contagions, instability”, the report cautioned.

The largest study into the subject of algorithmic trading, a 2014 paper that analysed data from 42 stock exchanges, found it makes markets more liquid and efficient but also increases short-term volatility. This is not a case of “good” volatility that arises from faster price discovery, the researchers added.

In fact, it appears HFT can lead to more volatility because markets are increasingly responding to the price changes triggered by computer algorithms rather than to new fundamental information.

That price changes can trigger further price changes is not a new phenomenon. A wealth of research confirms momentum has long been one of the primary drivers of stock returns. A 1999 paper found about 30 per cent of traders focused more on technical analysis – the study of charts and market prices – than on fundamental information. Indeed, one famous study, What Moves Stock Prices?, looked at the 50 largest one-day US market moves over the 1946-1987 period, and concluded many of the biggest movements could not be linked to obvious fundamental shocks.

Self-generated activities Clearly, it’s fanciful to think that before HFT, markets were driven by fundamentals alone.

However, markets are increasingly being driven by “self-generated activities” rather than by fundamental developments, according to a 2013 study of commodity prices. It found reflexive trading – that is, trading catalysed by price changes as opposed to by changes in market fundamentals – now accounts for between 60 and 70 per cent of price moves in the main commodity markets. Prior to 2005, the paper found, reflexive trading accounted for less than 40 per cent of price moves.

The authors, who include Prof Didier Sornette, a physicist renowned for his work on the development of market bubbles and crashes, investigated various reasons for the pronounced change in market behaviour, ranging from "behavioural mechanisms and herding" to hedging strategies, leveraged trading and the use of stop-loss orders, among others. However, the most credible explanation is the development of algorithmic trading "and of high-frequency trading in particular", they conclude.

Excess liquidity Prof

Sornette is not convinced by the argument that markets benefit by the increased liquidity provided by HFT firms. Excess liquidity is not always a good thing, he cautions in another paper, Crashes and High-Frequency Trading.

It can, he says, increase the risk of herding, “possibly leading to systemic instabilities and ultimately to crashes and their aftermath.”

Not only that, HFT traders provide liquidity “in good times when it is perhaps least needed” only to steer clear in times of market trouble, thereby adding to rather than mitigating instability.

Certainly, that appears to have happened during the 2010 flash crash. Stocks had been declining throughout the day; as volatility increased, the algorithms decided it was time to exit the market. With no one offering to trade, prices went into freefall.

Consulting firm Accenture, for example, saw its share price collapse from more than $40 to as low as one cent.

Investors should expect repeat episodes, concludes Sornette, who says high- frequency trading has caused crashes in the past, “and it can be expected to do so more and more in the future”. He also predicts the development of “crash algorithms” programmed to profit from market turbulence.

Sornette’s vision may be an overly pessimistic one. High-frequency trading’s heyday may already have passed; profits have declined in recent years and regulators are paying increased attention to the activities of HFT firms.

Mini flash crashes Still, computerised trading is here to stay, so some market hiccups can be assumed.

On one day last October, for example, there were 179 "mini flash crashes" – brief, dramatic drops in individual stocks – during the first 15 minutes of US trading, according to data firm Nanex.

What are the takeaways for investors? Firstly, it might be wise to avoid the use of stop-loss and market orders, and instead stick to limit orders that will be filled only at a specific price.

Secondly, investors should take with a pinch of salt media narratives that try to explain away daily market movements by exclusive reference to some news story or another. Rather than being catalysed by fundamental developments, such price changes may simply be the product of some meaningless algorithmic sparring.

Read More