Sharp fall in Nifty: Understanding flash crash, algo trading

http://timesofindia.indiatimes.com/business/india-business/Sharp-fall-in-Nifty-Understanding-flash-crash-algo-trading/articleshow/16702041.cms

Global markets have witnessed flash crashes in the recent past, the most famous one being on May 6, 2010 when US markets dropped 600 points in a matter of minutes, only to recover later. But the one which we witnessed Friday on the National Stock Exchange resulted in the market being shut for a while. 

Generally, algorithmic (algo) trading is associated with flash crashes as unlike a manual trader, the software logic doesn't have the mind of a human. With several algos at work across various brokerages, one such abnormal move activates the programme logic at umpteen algo desks, resulting in destructive market moves. 

Any manual trading terminal having such huge limits are manned by experienced dealers with enough safeguards, hence I was a bit flabbergasted when it was reported that the NSE flash crash was due to manual error and not due to algo-related orders. 

This is how an algo flash crash works: Assume there is a set of algos across unconnected desks which are programmed for directional trade, whether up or down. As long as one is making profits, the programme rides the trade. Once an abnormal trade (generally in frontline index stocks) is entered into the system by any broker and a specific stock or a set of stocks start moving down sharply, a number of other algos come into play based on their programmed logic, and thus go short on the specific stocks or the index, as the direction is down. 

And mind you, this happens in a matter of seconds or micro-seconds because that is how fast the orders flow out of the algos system. 

Thus a system which in a normal market provides liquidity and depth becomes a monster in no time - much like Rajnikant's Robot. 

Before anyone realises what's happening, the market could crash in seconds. 

SEBI has issued guidelines and safeguards for algo trading but the question is whether we have the wherewithal to ensure it is followed in letter and spirit. 

The Delhi high court is slated to hear a petition on January 29, 2013 for a ban on algo trading. 

Algos do add depth to the market and also contribute to the revenues of the exchanges due to the huge turnover, but it's still a domain of the chosen few. 

It's a different matter that algos have made a number of market participants redundant, especially jobbers and arbitrageurs. But who bothers about them as long as the cash registers are ringing.