The Securities and Exchange Board of India (Sebi) is working on rules to regulate algorithmic trading effectively.
According to two sources familiar with the matter, among the points under consideration are means to slow down the pace of trading through introduction of measures, including a minimum resting time for orders before execution, and randomising the time priority of orders an exchange receives.
Algorithmic trading refers to the use of electronic systems, which can potentially execute thousands of orders on the stock exchange in less than a second.
The regulator has been holding discussions on rules which could help create a more level playing field for non-algo players too, by placing some restrictions on how fast algorithms can trade in India.
"In the past, Sebi has consulted exchanges regarding concerns relating to fairness, transparency and equal access in view of increased usage of algo trading, co-location facility of the stock exchanges. There is thought within the regulatory body to have more checks in place," said a source.
To eliminate "fleeting orders" or those that appear and then disappear within a short period, a mechanism might be introduced to prevent cancellation or modification of an order until some time from its submission.
"We are mulling to introduce a minimum resting period of 500-600 milliseconds," said a source.
The introduction of a minimum resting time could help eliminate the potential for misuse inherent in 'fleeting' orders, said the technology in-charge at one of the top five brokerages.
Such orders, which are created and cancelled in short periods can create the impression of artificial liquidity and affect the behaviour of the rest of the market. This can have an impact akin to price manipulation. Introducing a measure to eliminate such a possibility would be a positive said the person, on condition of anonymity.
The regulator is also looking at randomising all orders received within a set period, say within two seconds. All the orders received within that period would arrive at the exchange in a random manner.
Those in the algorithmic trading space say such moves would have a significant impact on operations.
Hitesh Hakani, director of Greeksoft Technologies, said randomisation of orders could severely impact the utility of colocation servers. These servers are located within the exchange premises to execute trades faster by shaving off the fractions of a second that an order takes to travel to the exchange.
"There would be no point in availing of colocation servers if there is to be randomisation between colocation and noncolation orders...These moves would have a major impact on algorithmic trading players who are now significant liquidity providers in the markets. Every millisecond counts for them," he said.
Under the rule of randomisation, orders received during a period (say one to two seconds) could be modified to have a new priority and sequencing.
"Such a move would slow down algorithmic trading engines and make redundant existing systems such as co-location and high speed networks. They would serve little purpose if orders are to be slowed down and randomised," said Naveen Kumar, founder and chief executive officer at QuantXpress Technologies.
Non-algo players see some of the moves as positives.
Says A Balakrishnan, chief technology officer at Geojit BNP Paribas Financial Services, "They are building latency. Often when an opportunity arises in the market, algo traders can take advantage of it faster than manual ones. This would look to bring them more on par with each other."
He estimated it takes around 300-400 milliseconds for a retail investor's order to get processed. The introduction of a resting time of 500 milliseconds or more could be to bring the latency on a par with that of a non-algorithmic trader, according to him. This new moves if implemented could also create issues for algorithms, which trade in different markets.
"If there is an institution, which invests across different markets, for example in India and Singapore, then they will have to recalibrate their systems if latency is going to be introduced into the market," said Balakrishnan.
The bulk of trades on the National Stock Exchange and the BSE happen via the algo route. As of May, 15.09 per cent of the BSE turnover came through the algo route. Algorithmic trading accounts for 19.83 per cent on the National Stock Exchange, with an additional 24.26 per cent of volumes coming in through co-located servers.