2013-08-28

Attack of the Trading Machines

The Day Everbright Had to Pull Power Plugs to Stop Erroneous Trading
Everbright has suspended Yang Jianbo, head of its Strategic Investment Division (SID), which is directly responsible for the mistake. Soon after, Xu Haoming resigned as the firm's president.

Founded in 2010, the SID was a rising star in Everbright. It earned 124 million yuan in profit for the company in 2012, a 33-fold increase from the previous year.

The glory ended on August 16. Within two seconds at 11:05 a.m., its trading system generated 26,082 buy orders, sweeping clean all stocks for sale on the main board and pushing up the Shanghai Composite Index by 5.62 percent from its closing level the previous day.
It happened because the order-generating system was programmed to resend orders if it did not receive feedback from the order execution system in 150 seconds, an Everbright executive said.

In itself, the repetitive operations are unlikely to cause much trouble because they need to pass through layers of examination and risk control before being sent to the stock exchange and executed.

But all of these layers failed. By the time the morning trading session closed, Everbright had offered to buy stocks worth 23.4 billion yuan, and shares worth 7.3 billion yuan were actually bought. Traders canceled the rest, resold some of the stocks and shorted the stock index future, narrowing the firm's risk exposure by the end of the day to 196 million yuan.

When the traders realized the system had sent out a huge amount of unintended orders, "new orders were still being generated and there was no way to stop them," a source with firsthand knowledge of the situation said.

"The traders had to pull out the Internet cable and also the power cord" to shut down the computers, he said.
The article goes on to place some blame on the Shanghai Exchange as well. In any case, these types of events are always possible, but the odds of them occurring are more likely during periods of negative social mood. More importantly, the effect is far larger during negative mood. During positive social mood, these types of blunders would mainly be a joke to traders and investors, but during negative social mood, it makes them question the stability of the system.

No comments:

Post a Comment