In the past few years, financial exchanges have experienced several incidents of “flash crashes” and a few “flash booms” as well. These are instances when prices shift huge amounts within seconds. They are caused when computers execute trades on massive volumes, entirely without human intervention and according to pre-programmed algorithms. The term was coined following the most notorious of flash crashes, on May 6, 2010 when the Dow Jones Industrial Average dropped nine per cent in a minute. The latest incidents occurred last Friday. Nifty futures dropped 6.7 per cent in a minute as 35,000 contracts were offered (1 contract = 50 Nifty) at a large discount. Earlier that same day, Infosys stock futures also dropped 20 per cent for an instant. On March 22, the world's most valuable company, Apple, saw an instantaneous drop of 9.5 per cent in trading on the US' BATS exchange. The stock of BATS itself has suffered twice from similar incidents.
These events occur with sufficient regularity to present a serious problem, and the incidence is likely to rise. High-frequency trading (HFT) via algorithms generates well over 50 per cent of global volumes and its popularity is growing. The danger is that the flash crashes linked to HFT could cause real contagion and panic. One computer sells a big order at a big discount, maybe in error; that triggers responses from other computers, which are programmed to execute automatic stop-loss orders if a given price is hit. This nightmare occurred as long ago as Black Monday, October 19, 1987, when NYSE traders returned from lunch to see the Dow down 20 per cent.
Luckily, contagion didn't happen on Friday. Both the Nifty and Infosys recovered quickly from the lows. But it could be just a matter of time. HFT is becoming more popular in India. Over 100 traders already rent terminal space on the NSE premises, hooking directly to the trading servers. The milliseconds saved by “co-location”, as this is called, translate into big profits. Apart from legitimate arbitrage, a co-located terminal enables some sharp practices. Programmes can spit out a stream of opposed buy-sell orders on the same underlyings to determine liquidity across a wide price range; other programmes stretch ethics even further by issuing orders and cancelling them in seconds solely to determine if these are matched. Regulators recognise the problems, but have never found foolproof safeguards against the dangers of HFT. Experiments with circuit-breakers and stress tests to determine likely contagion limits have not been completely successful. Short of drastic solutions like outlawing HFT itself (at the cost of huge volumes), there may be no silver bullet. However, stop-gap measures like a mandatory human confirmation of an unusual order could reduce the chances of contagion. Sharp practices also need to be identified, outlawed and penalised ruthlessly. Market regulator Sebi can count itself lucky that Friday’s flash crashes did not occur in the last minutes of the session. That left time for a recovery. It's good to hear that the market regulator is investigating what exactly happened and reportedly setting up a series of stress tests. It must not delay the issue of revised guidelines for HFT.