India has had its share of mini flash crashes. Earlier this year, Nifty futures fell by around 6% in seconds, thanks to an erroneous order which resulted in freak trades. Friday’s flash crash was of a much greater degree, dragging the Nifty index lower by 15.5%, again in a matter of seconds. This time as well, the source of the problem was an error from an institutional broker, Emkay Global Financial Services, although there were a number of wrong orders in Nifty stocks, which dragged the index lower. These orders, also known as “fat finger” orders, resulted in trades worth as much as Rs.650 crore.
The losses for investors, the institutional broker and its client would have been larger, but for the Sebi-mandated circuit breakers that apply when the markets drop sharply. Since all of this damage happened in a matter of seconds, it seems reasonable to assume that this was caused by an erroneous algorithmic trading order. But the National Stock Exchange has clarified that the problem orders weren’t generated by an algorithm—the exchange will know because orders are tagged as algo if they originate from a software-driven terminal.
In both of the above-mentioned flash crashes, therefore, the source of the problem was a manual error, rather than an algorithmic error. Of course, it’s fair to say here that even when an algorithm malfunctions, it’s because of a manual error or oversight—either the software may not have been tested adequately, or sufficient risk controls may not be in place.
Having said that, in today’s fast-paced trading world, when losses can mount very quickly, it’s imperative that regulators and exchanges put in place controls to minimize losses. In India, the regulator has a pre-trade order filter of 20% for stocks that are part of the Nifty and the Sensex. As a result, traders can place orders in these stocks at a price that is +/- 20% compared with the previous day’s close.
One way to minimize losses is to narrow this band. Needless to say, this will be viewed negatively by many traders, who want to be as nimble as possible in a fast-changing market. But it’s possible to make the price band a moving target, where traders don’t have a large 20% price band to start with, but perhaps a 10% band, which moves along with the volume-weighted average price in the preceding 30 minutes. In any case, it’s difficult to see why any trader would want to suddenly enter an order that’s 20% higher/lower than prevailing prices.
Earlier in the year, as part of its revised guidelines for algorithmic trading, the Securities and Exchange Board of India (Sebi) had said that broking firms should put in place pre-trade filters to ensure their errors don’t affect the entire market. But the regulator stopped short of imposing a strong penalty for defaulters. If there is a harsh punitive action in place, broking firms will be forced to ensure tight risk controls, which will again go a long way in minimizing risks.