Nasdaq OMX connectivity disaster highlights stumbling markets
A three-hour trading crash at Nasdaq OMX caused by a connectivity issue has once again put the spotlight on trading technology and the resilience of financial markets, which have been sorely tested in recent months and years.
On Thursday afternoon, Nasdaq OMX realised that price quotes were not being disseminated by the securities industry processor (SIP), which consolidates and disseminates all prices for the industry. The problem was traced to a connectivity issue between an exchange participant and the SIP, which meant the SIP was unable to disseminate consolidated quotes and trades.
Nasdaq chose to halt all trading until the issue could be fixed – a process that the exchange says took 30 minutes. The remainder of the time was spent coordinating with regulators and other exchanges to ensure the market worked properly when reopened. However, shares fell by 3.39% after trading resumed.
“Nasdaq OMX will work with other exchanges that are members of the SIP to investigate the issues of today, and we will support any necessary steps to enhance the platform,” said the exchange in an official statement.
The incident follows a spate of trading crashes and disasters as Nasdaq OMX, other exchanges, and major brokers. Nasdaq was fined $10 million in May for failures relating to its disastrous Facebook IPO last year, while an error in its data feeds in January caused an outage that left investors unable to view Nasdaq-listed stocks.
However, Nasdaq OMX is far from alone in its troubles. Rival exchange BATS Global Markets faced a humiliating debacle over its own botched IPO in March 2012 and on 10 January, BATS admitted that it had uncovered a technical fault that caused it to accidentally breach best execution regulation on thousands of client transactions over a four-year period.
The disasters have spread to brokers and major bulge-bracket banks, too. Earlier in August, Goldman Sachs suffered a trading error that cost the firm an estimated $100 million. That incident was caused by a computer error in which the bank’s automated trading systems accidentally sent indications of interest as real orders to be filled at the exchanges, resulting in a number of erroneous options trades that disrupted trading across US exchanges on 20 August.
Meanwhile a few days earlier, a trading error at broker Everbright Securities in China on 16 August caused the Shanghai stock market to experience a sudden 53% surge in volumes that rattled Asian markets. The incidents at Goldman Sachs and Everbright Securities follow on from similar high-profile cases last year. In August 2012 broker Knight Capital suffered a catastrophe in which a problem with its high-frequency market making software caused a loss of $440 million that forced the firm to recapitalise and seek new backers. Knight eventually merged with rival Getco four months later.
Automated trading in general has been in the spotlight ever since the flash crash of May 2010, in which the US stock market unexpectedly plummeted by $1 trillion, then just as quickly rebounded. That event has been debated ever since, with many placing the blame on interlocking algorithmic systems feeding off each other in a race to the bottom after an accidental fat finger trade. According to some observers, the failure of sophisticated trading systems has become sufficiently commonplace to suggest there is an underlying problem in the culture surrounding sophisticated trading strategies.
“The pattern of technical problems that we are seeing is definitely undermining confidence of investors in having fair access to the markets,” said Lev Lesokhin, executive vice president, strategy and market development at Cast Software. “Most IT applications have dead code. It’s in there just hanging out in the code base but none of the live modules are calling it. If you don’t have structural oversight then you don’t know if your new live code could be calling the dead code. In the case of Knight Capital’s outage last October, the live code called the dead code back to life and the program started trading on that – at Knight, that bizarre thing was Frankenstein code.
“This problem is the result of hundreds of developers working every day, changing code, putting it together and testing it. The way that process is overseen and the construction and the quality of the construction is something that the industry really needs to catch up on.”
Some countries have attempted to bring in legislation to combat the problem. In Germany, firms practicing HFT have to register with the authorities, and provide details of their algorithms and how they operate on demand. At the European level, the EC’s upcoming MiFID II legislation contains similar proposals to ensure adequate testing, maintenance and regulatory oversight of algos, which were issued as non-binding automated trading guidelines as far back as February 2012.