Shyamal Banerjee/Mint
Shyamal Banerjee/Mint

Is algorithmic trading over or under-regulated?

The algo trading landscape is complex; regulations must continue to evolve along with the segment

Two years have passed since the Securities and Exchange Board of India (Sebi) issued a contentious discussion paper on co-location facilities offered by stock exchanges, but there’s no news yet of any new regulations on this front. In fact, Sebi has largely been silent on algorithmic trading for some time now.

One would have imagined that the trading community was pleased with the state of affairs, especially given Sebi’s inclination to somewhat restrict algorithmic trading (see http://mintne.ws/1acWu7h, for instance). But one of the thoughts that emerged at a FIX Trading Community conference last week was that the markets will be better off if Sebi proactively issued new guidelines for algorithmic trading, rather than clamp down in response to a trading accident such as a freak trade.

Be that as it may, in some cases, the opposite also holds true—a number of traders were miffed about Sebi’s mandate for testing changes in trading software at a special mock session conducted by stock exchanges. There appears to be, by the trading community’s own admission, a curious mix of over-regulation and under-regulation in the algorithmic space.

For perspective, Sebi had issued broad guidelines for algorithmic trading in March 2012, some months after an erroneous algorithm created havoc during BSE’s Muhurat trading session. But trading accidents continued, such as when the price of Infosys Ltd futures fell by 20% on the National Stock Exchange (NSE) due to an error in an algorithmic trading software. This was followed by the freak trade by Emkay Global Financial Services Ltd, which forced a brief trading halt on the NSE, although this was a so-called fat finger error and had nothing to do with algorithmic trading.

Not surprisingly, Sebi tightened some algorithmic trading norms and also issued guidelines for testing changes in trading software in 2013. While much has changed since the disruptions in 2011 and 2012, the fact remains that the markets aren’t accident-proof. As such, one of the worries is that another major disruption may cause the regulator to clamp down in a hasty response.

In response to the Emkay episode, stock exchanges had rather quickly reduced price limits on individual stocks. As pointed out in this column back then, the drawback of this move was that it would end up hurting the depth of the markets further, as investors would be able to place orders only within a restricted band. What the Indian markets really needs is a limit on order execution, rather than tinkering with the order book.

Also, the new system has not been really tested by an algorithm going awry. According to the regional head of electronic trading at a multinational broker, “The current system doesn’t preclude a trading accident from disruption of the markets, and hence Sebi must proactively put in place a better system for preventing or minimizing market disruptions." About a year ago, NSE adopted a version of the US ‘limit up-limit down’ mechanism, which is designed to prevent trades from occurring outside of a specified price band, and can be an effective tool for protecting the market from sharp movements caused by erroneous orders. It had introduced this only for the mid-month and far-month contracts in the derivatives segment, where the lack of liquidity makes it difficult to assess how effective the change in market design has been.

Sebi, however, appears content with the way things are functioning and seems the least bit interested in tinkering with market design. It would do well to borrow caution from one section of the trading community—the algorithmic trading space is a fast evolving one, and the regulator must always be on the top of the game. The only area in the regulator’s radar appears to be guidelines for co-location; hopefully by the time these are issued, Sebi would have scrapped the strange proposal to implement a new order handling system, with two separate queues for co-located and non co-located orders, such that the order matching engine picks from each queue alternatively.

Co-located facilities enable members to set up automated trading systems in the same building as the exchange, to reduce latency or the time required for data flow between the exchange and the broker’s trading system. Sebi’s attempt to bridge the gap for those who can’t afford co-location is bizarre, to say the least. Those with more financial resources will always be better off than retail investors, both in terms of speed of access and market research. Such factors must not be regulated, and point to the regulator’s propensity to over-regulate in some areas. At the same time, some guidelines will help rein in anti-competitive practices by exchanges.

Similarly, representatives of trading firms say that the guidelines for testing changes in algorithmic software are not only cumbersome but can also end up increasing the risks to the market, rather than reducing them. Besides, while one stock exchange is overly cautious in approving algorithms and changes in software, the other isn’t as particular. Evidently, in the eyes of the former, the latter’s approach will pose risks to the system. But in any case, it’s a bit much to expect for-profit stock exchanges to eliminate all risks and make market integrity their prime concern. Ironically, then, in the area of testing of software, the current approach comes across as one of over-regulation; even though the apparent chinks in the system calls for a fresh set of guidelines.

All told, the algorithmic trading landscape is a complex one, and Sebi’s regulations must continue to evolve along with the segment.

We welcome your comments at inthemoney@livemint.com.

Close