For instance in late September, news broke that Apple would release its own Internet radio service, a move that was widely expected to deal a severe blow to rival services like Pandora and Sirius.
At 3:30 p.m. that day, Pandora's stock dropped from $9.50 to $7.50 based on a hot headline, according to Lobravico, VP of risk arbitrage, trading and sales for Wall Street Access, an institutional broker. The stock then climbed back to $8.50.
Lobravico says he also sees similar moves in after-hours trading.
Algorithms Also Becoming A Worry?
Traders are also worried about algorithms becoming more aggressive.
"Newer, faster, dirtier, nastier algos are coming out every day. I think it's kind of a natural progression," Lobravico says. "The pioneers with the algos that came into the market 10 years ago are much different than the ones they have today."
Algorithms are programmed to execute when certain parameters are met. "A lot of times you're getting an erroneous trade, a fat finger or legitimate news story or something that sparks a rumor in the stock, which triggers another set of parameters," he explains.
For instance, on Oct. 5 a single algorithm was responsible for 4% of U.S. stock-quote traffic, according to Nanex. The program executed at 25 millisecond bursts in 500 stocks after the market's open, and then abruptly stopped working at 10:30 a.m., Nanex says.
Commenting on the incident in a Forbes blog post titled "The Virgin Slut Algorithm," David Leinweber said the algo ate up 10% of the communications capacity for the entire marketplace.
So why the provocative title? The "virgin" part is that the algo "never executed a single trade" and canceled every order, he wrote.
KOR's Nagy attributes such events to the interconnected nature of the markets and the fact that algorithms are listening to each other. "They're all following each other's footprints, time and sales. They're all listening to the futures market," Nagy says. "If a stock is dropping, the algorithm stops liquidity. It's not like you can force these algorithms to participate in the marketplace."
Because of that, this sort of problem is not easily solved, while the list of mini-flash crash incidents continue to grow.
On Oct. 3, an error sent shares of Kraft Foods soaring 29% during the market's opening minutes, although the trades were later canceled. Kraft had recently switched its listing from the New York Stock Exchange to Nasdaq after the company split itself into two units.
Initial reports that day attributed the faulty price spike to a technical glitch or errant algorithm, but Nasdaq later said it was broker error.
Following the 2010 flash crash, exchanges sought to restore investor confidence with circuit breakers by reducing the potential for excessive volatility. They also strengthened the "clearly erroneous trade process" to alleviate the damage that could be caused by fat finger errors. In such instances, a broker-dealer can have errant trades reviewed by the exchange. In turn, the exchange can either break the trade or force the broker stand by it. But what exactly qualifies as an erroneous trade?
If a trader who enters a 5 million-share order intends to use a volume-weighted average price algorithm that executes between 11 a.m. and 4 p.m., but accidentally sends an order that will significantly move the price, the exchange will usually uphold the transaction, according to Wall Street Access' Lobravico.
The only way an exchange will break a trade is if it's a legitimate error, he says. "If a trader types in an erroneous price that is 30% to 40% away from where the stock is, as in the case of Kraft, that print should come down and any related trades that were affected by it will be broken as well," he explains.
But sources say that mini-flash crashes can occur within the bands of the circuit breakers, so they are not always detected. Also, not everyone finds these trading halts to be helpful.
"We have these circuit breakers now. Sometimes you have orders out there and orders are halted and you don't know where you stand," says Lobravico.
"It can take Nasdaq a half an hour to find out what happened, and you get print cancels. You thought you were long or short X. That is a massive liability for broker-dealers who are representing customer orders, because they're out there trading in a market structure that is completely flawed," he says.
In terms of remedies or solutions to curb such episodes, Nagy suggests that the pace of rule-making needs to be slowed down. "We don't understand the unintended consequences of the interconnected marketplace," says Nagy. In particular, he says the pace of exchanges filing for new order types has accelerated, and he contends that it's created a number of mini flash crashes. The rule filings are all "one piece in a larger puzzle of the market structure," he asserts.
The inability of regulators to see all the data has been a hindrance to understanding the causes of the mini-flash crashes as well, according to several sources. "It goes back to the consolidated audit trail," which should give the SEC the ability "to see who is the first mover and who is reacting and interacting with it," Nagy says.
