Old-timers who are yet to acquaint themselves with the Internet and online trading view algos as an invasion from another sphere, out to destroy their cosy, familiar world.
These fears are not misplaced. Algo trading, where computer-driven software places and executes orders on behalf of an investor or trader, have already over-run stock markets; not just in India but across the globe. The National Stock Exchange (NSE), which accounts for over 80 per cent of the turnover in India, reports that almost 40 per cent of the trading in its cash segment is done through algos. In the derivative segment, algos account for a larger 44 per cent share. The numbers are higher in global exchanges in the UK and the US, with these trades estimated to account for over 60 per cent of turnover.
Back-door entry
The detractors have a reason to feel cheated because this trading method clearly gives its users an edge. Machines are, after all, faster than men. While it is perfectly legal and above-board, the entry of algo trades into Indian markets seems to have been made from the back door rather than the front. For instance, attempts to find the SEBI circular permitting the use of algos in India failed, because it does not exist.
In 2008 SEBI gave brokers permission to allow their larger clients to directly access the exchange server without intervention of the broker. By allowing direct market access (DMA), the regulator implicitly permitted the use of algo-driven programmes for trade execution.
In August 2009, the NSE started selling co-location space (colo), which offers brokers the facility to locate their servers in rack space close to the exchange server. Since the latency (time taken for data to travel from one point to another) depends on the distance from the exchange server, many wealthy investors bought such rack space so that their algos could be ahead of most others. As the usage of DMA and colos in India increased, so did the use of algos.
Not all that bad
So algos came, became popular and are here to stay. There are complaints that some algos are flooding the exchange with orders, many of which are eventually cancelled. This is because some algo software resort to High Frequency Trading (HFT), wherein they fire multiple orders at high speed to the exchange server. The NSE has, however, tried to address this issue by imposing a penalty on traders with high order-to-execution ratios. Some also complain that the algos perched in co-location facilities are the first in the queue and thus get the best price for their trades.
While these complaints are true, some good has come out of them. For instance, market volume has grown in the derivative segment after the advent of algo trading. The bid-ask spread, or the difference between the buy and sell quote for a stock, has narrowed considerably due to the increase in volumes.
How to join in
The good news is that algos are no longer the preserve of the rich. Anyone who trades from home through online trading can use algos too. But only plain vanilla algos are allowed in the Indian markets. The rogue algos, such as those that go snooping into other computers or scan twitter messages to put in trades, are not allowed by SEBI.
In the cash market in India, algos can be used to execute trades closest to volume weighted average price (VWAP) or time weighted average price (TWAP). That is, an investor can use these algos to put in buy or sell orders at rates where the maximum volume was transacted or when the volume was the greatest. These programs can be used by large institutional investors too, to buy and sell at optimal rates. It reduces the burden on the dealer who has to constantly monitor stock prices to execute a large order.
Some algos compare rates across exchanges and put in the trade in the exchange that has the most favourable rate. The popular algos in the derivative segment spot price differences between the cash and futures price of a security, or the calendar spreads of derivatives, and execute trades based on these anomalies.
The costs
The cost that a trader or an investor incurs varies according to the needs. Strategies that are latency (or time) sensitive will work only if rack space is purchased on the exchange. The running cost can then be ₹30-40 lakh a year. But if the strategies don't depend on speed, the algos can be parked on the broker’s server setup or even on your personal machine. The running costs can then be ₹3,000-30,000 a month or more, depending on the setup.
There are specialised software by Flex Trader, Omnesys/Thomson Reuters, among others. But a strategy can also be designed on an excel sheet and automated. You can even ask your broker if he can provide the application programming interface to connect your Excel to the broker’s execution management system to automate your program.