Recent flash crashes on the Indian and overseas bourses tend to indicate that technology-enabled speed-trading on the bourses, at times, is no boon.

There are instances of exchanges annulling trades and in some cases not doing so, after investigation.

While the frequency of these momentary lapses, triggered either by man or technology is not too alarming, the fact remains that such errors cause losses.

Last week, Reserve Bank of India, in its financial stability report, said a proper regulatory structure and continuous monitoring of regulatory systems would avert operational and other risks posed by algorithmic trading (Algo) and high frequency trading (HFT).

HFT uses computer algorithms to analyse market data, deploy trading strategies, minimise cost of trading and execute trades.

The algorithms interpret market signals and patterns, and trades are executed and reversed within a fraction of a second, multiple times during the trading session.

SPEED

The Bombay Stock Exchange is capable of handling one lakh orders a second.

“We get about eight crore orders in a six-hour session,” said Ashish Kumar Chauhan, Managing Director and Chief Executive Officer, BSE.

Latency on the BSE is below 10 milli-seconds, while the world has moved to 100 micro-seconds.

Latency is the time taken for stock prices to be disseminated from the exchanges’ IT network to the broker terminals.

It is measured in seconds. (1,000 milli seconds make one second and a million micro seconds make a second).

The flash crash on the NSE (October 5, 2012) sent the Nifty plunging 900 points, from 5787.6 to 4888.2, due to 59 erroneous orders keyed in by a broking house. This resulted in multiple trades (non-algo), totaling Rs 650 crore. The incident occurred at 9:50.58, triggered the Nifty circuit filter, upon which the cash market closed automatically. The market resumed at 10:05 a.m.

The NSE acted immediately and the defaulting broker was made to close out his position (equal and opposite transaction at current market price) and then disabled from trading.

Closing out means dealers sell or buy the quantities bought or sold, which ensures that loss or profit on this count is on his account.

On April 20, 2012, NSE’s Nifty futures contract for April saw a freak trade that valued the contract at 5,000 (310 points down from opening price).

Earlier on the same day, Infosys stock futures trading at Rs 2,400 suddenly lost Rs 450 and touched the day’s low of Rs 1,950.

NSE stated that its trading systems worked normally and all the trade executions were within the price limits prescribed by SEBI. Marketmen said the dealer who routed the trades would have lost Rs 7-8 crore on the Nifty and about Rs 1-2 crore on the Infy scrip alone.

The BSE annulled derivative trades done on October 26, 2011, as it detected a “large movement of Sensex futures.” There have been similar instances across the globe, especially where algorithm-based trading strategies are in place. On May 6, 2010, major equity indices on the Dow Jones suddenly fell 5- 6 per cent and recovered within minutes. About 8,000 stocks and exchange traded funds lost 5-15 per cent and recovered most of their losses. Over 20,000 trades across 300-plus scrips were executed at prices that were over 60 per cent away from their value moments ago.

A joint report by the US securities and the commodities market regulators said liquidity could rapidly erode and result in disorderly markets due to automated execution of algorithm-trading strategies under stressed market conditions. In April, the International Organisation of Securities Commissions, a policy forum for securities regulators, came out with a list of recommendations to strengthen surveillance capabilities of regulators to minimise such risks.

> shanker.s@thehindu.co.in

> raghavendrarao.k@thehindu.co.in