Join our community of smart investors

Rise of the machines

High-frequency and algorithmic trading bring advantages. But they can also cause flash crashes. Jonas Crosland reports on a proposed solution
November 2, 2012

There was nothing unusual about the start of trading in New York on 6 May 2010. But worries about the Greek debt crisis continued to bubble to the surface and, by 2.42pm, the Dow Jones index was down more than 300 points. However, in the next five minutes it dropped another 600 points, and at one stage was down 998 points - the largest one-day drop ever.

In that brief time, more than $1,000bn was wiped off the value of shares and, as the dust settled, over 20,000 trades were cancelled. Institutions created the mess, and private investors were naturally left wondering what had happened. The more lasting effect was that many of those private investors withdrew from the equity market for good.

So what did happen? There were several factors that made a contribution, but the main catalyst was a mutual fund placing an unusually large sell order on an S&P 500 futures transaction valued at around $4.1bn, basically as a hedge on an existing equity position. This happened against a background of negative sentiment and thinning liquidity, and triggered a frenzy of selling from trading firms specialising in high-speed trading.

The key point here is that a joint report from the Securities and Exchange Commission and the Commodity Futures Trading Commission took nearly five months to analyse what happened in just five minutes of trading.

 

 

Regulation struggling to keep up

Clearly the advances made by using high-frequency trading and computer-driven algorithms have been far greater than attempts by the regulators to keep up by providing sufficient checks and balances. And such measures are being looked at by the UK government's Office for Science, which has come out with a range of suggestions that might help to avoid a repeat of the 'Flash crash' in 2010.

Such an investigation is especially pertinent given that nearly two-thirds of all transactions in the US and nearly a third in the UK are now conducted using high-frequency computer-based trading.

However, its research has found that despite the widely held negative perceptions, high-frequency trading (HFT) and algorithmic trading (AT) have brought advantages such as improved liquidity and lower transaction costs both for retail and institutional traders. But in certain circumstances computer-based trading can lead to significant instability. The report particularly highlights self-reinforcing feedback loops, where a potentially minor piece of information is fed back into investment models, thus generating an ever increasing spiral of computer-related trades.

But the effects are not intentional, as no-one benefits from a market that loses all pretence of being orderly, and while there is no evidence to confirm that there is any market abuse, there have been claims of market manipulation.

Three of the areas highlighted include suggestions that high-frequency traders exploit their speed advantage to disadvantage other participants in financial terms. HFT traders might also engage in market abuse, which is distinct from securities fraud, by using predatory tactics such as bluffing by appearing to be informed of some key information, or other methods of price distortion.

 

 

Changed for good

There are other more fundamental issues, too. HFT relies on high-speed computed linkages, and positions are rarely held for long - sometimes just microseconds - and some traders operate with very low levels of capital, while traditional traders have an obligation to stand ready to buy and sell.

Moreover, HFT traders manage their risk by stopping trading when conditions are not ideal, which means that liquidity levels can dry up very quickly. Economist and columnist Paul Krugman commented: "The stock market is supposed to allocate capital to its most productive uses, for example by helping companies with good ideas raise money. But it's hard to see how traders who place their orders one-thirtieth of a second faster than anyone else do anything to improve that social function."

However, the report makes it clear that to walk in with a large legislative stick runs the risk of killing off some of the advantages that computer-based trading brings to the market. And this kind of trading is set to grow further. Far fewer human traders will be needed in major financial markets of the future, while computer-driven trading will become ever more complex.

 

 

Keeping technology under control

In an attempt to establish a greater level of supervision in this fast-changing market without creating a regulatory burden, the reports suggests a wider use of circuit breakers. These can take many forms, including halting trading in a particular stock or an entire market, or by setting limits on price movements.

The London Stock Exchange already operates a stock-by-stock circuit breaker that, when triggered, operates in auction mode, whereby an indicative price is constantly posted while orders build up on either side. At some point the auction is withdrawn and normal trading resumes. This is no cure all, though. Companies issuing profit warnings should see the share price fall to reflect this, and placing an artificial brake on what is a normal consequence of the market working could be counter-productive. In other words, circuit breakers cannot forestall valuation changes that are unavoidable.

But, applied selectively, they could be used to rule out some of the worst consequences generated by order imbalances at a time of illiquidity, by introducing a cooling off period. So, when a large sell order comes along, all selling orders are held until sufficient liquidity is generated by counterparties who have been given time to understand what is happening.

Other measures centre on greater standardisation of electronic financial systems and improved standardisation of connectivity to trading platforms. But all of these proposals inevitably hit one very large brick wall. In order to be effective, there must be little short of global application. There is no point in shutting down trading in a stock in London when trading continues in New York or Singapore.

And that presents a tough challenge, given that global financial authorities can rarely agree on the time of day, let alone a package of complex checks and balances designed to cocoon financial markets from their own worst excesses.