For all the talk that computers are smarter than human beings, the fact remains that computer-driven algorithmic trading has also led to some bad experiences for market participants. It’s true that human traders are also prone to errors such as fat-finger trades, but there is ample empirical evidence that faulty algorithms can put the entire market at risk.

Shyamal Banerjee/Mint

In this backdrop, it was interesting to note some of the comments made at the recent TradeTech India 2012 conference, an annual meet held for users and vendors of electronic trading services. A broker offering electronic trading services said during a panel discussion that there is a need for greater regulation of risk management systems of firms that engage in electronic trading. It’s rather rare to hear a regulated entity making a call for greater regulation. But it must be noted that firms have to be prepared for more regulation, whether they like it or not. This is because regulators across the world are worried about the risks posed by algorithmic trading to the markets.

Also, the broker’s comments are nuanced. Currently, some countries, including India, follow the practice of checking and approving each and every algorithm that proprietary trading firms and brokers want to deploy. Most market participants disapprove of this practice since it slows the deployment of new algorithms and also involves revealing trade secrets. If regulators get involved in only checking the risk controls that are in place in a firm’s order management system, it will alleviate the above-mentioned problems.

Says Hariharan Sundaresan, head of proprietary trading, o3 Capital Global Advisory Pvt. Ltd, “There are no easy solutions for the regulator and the exchanges to manage the risk that arises from faulty algorithms. On the one hand, if the regulator and the exchanges make regulations very stringent, it will hit innovation in the industry. On the other, if they get lax with regulatory requirements, the entire system will be at risk. They must tread a fine line that enables firms to innovate freely, while at the same time protecting the markets against systemic risk.”

It’s also important to note here that the most innocuous of algorithms can have adverse effects if the right risk controls are not in place. It’s not only high frequency trading firms or firms that use complex algorithms that pose risks to the market place. The above-mentioned broker said on the panel that even the misuse of a vanilla VWAP algorithm (which executes an order at a security’s volume-weighted average price) can lead to a spike in the price of a security. Considering that there are a myriad number of firms that use electronic trading systems, it may not be all that practical to ensure that every firm has all the risk controls in place. Sure, firms that err on this count must be taken to task. But by the time the error is discovered, it may already have caused irreparable damage to the marketplace.

It’s imperative, therefore, that risk management at the exchange-end is taken much more seriously. Anant Jatia, founder director at Forefront Capital Management, says, ”The exchanges are the last line of defence. As such, they must determine what is the speed and reaction time they are comfortable handling, and design their risk management system accordingly. The right risk management framework at the exchange end will mitigate a lot of the risks to the system.”

Indian exchanges follow the good practice of setting position limits based on the margins deposited by a trading member, which safeguards the system to a large extent. Besides, the National Stock Exchange sets the throttle rate at a predetermined rate, which ensures that an algorithm that is faulty or is designed with mala fide intent isn’t able to pump in excessive quotes in a short period of time. Throttle rate is essentially the dealing capacity of firms, or the rate at which they can send orders to an exchange, and is measured on the basis of orders per second.

It must be noted here that an accident is possible despite having these safeguards. Proponents of electronic trading argue that big accidents are possible even with human traders. But having said that, the concern with electronic trading is that the speed with which algorithms execute trades will lead to large damages in no time, as was seen during the flash crash and the Muhurat trading incident.

It becomes important, therefore, to better understand the risks that are involved. JR Varma of the Indian Institute of Management-Ahmedabad and chairman of Sebi’s risk committee has suggested in the past that a stress test of the market structure should be conducted by building a so-called stock market simulator. This idea was first mentioned in early 2011 by University of Pennsylvania’s Michael Kearns, who was quoted in a Reuters column. Needless to say, the cost and effort involved in building such a system will be high. But at the same time it will be well worth the effort. At least, regulators can stop groping in the dark when it comes to their understanding of the risks involved as well as their regulation of algorithmic trading.

Your comments are welcome at inthemoney@livemint.com

Also Read |Mobis Philipose’s earlier articles