Two weeks ago, the Securities Exchange Commission and Commodity
Futures Trading Commission revealed the possible causes of the May 6th
Flash Crash. Since you can't stop traders from making bad trades, what's
to stop another Flash Crash from happening tomorrow? aiCIO's Joe Flood
Two weeks ago, the Securities Exchange Commission (SEC) and Commodity Futures Trading Commission (CFTC) released their heavily anticipated joint report on the causes of the May 6th Flash Crash. A team of 20 investigators headed by a former physicist and quantitative trader pored through reams of millisecond-by-millisecond trading timelines, consulted with academics and market researchers, interviewed dozens of high frequency traders, exchange operators and outspoken critics of each, and, according to insiders, did something perhaps even more remarkable: managed to create a feeling of equanimity and collaboration between two sprawling Washington bureaucracies with overlapping authorities and goals that are often at odds with one another.
The findings? A mutual fund from Kansas made a clumsy trade. At least, that seems to be the conventional wisdom in the media—and with their conclusion has come a hail of criticism: since you can't stop traders from making bad trades, what's to stop another Flash Crash from happening tomorrow? And now even that central conclusion—that one big trade triggered the collapse—is under attack from the market research firm Nanex, which has served as something of one-outfit opposition party to prevailing theories about the Flash Crash (proving itself right and garnering some advice-seeking attention from the SEC-CFTC investigators in the process).
First, the trade believed to be at the root of the crash. According to the SEC-CFTC report, “a mutual fund complex...initiated a sell program to sell a total of 75,000 E-Mini contracts (valued at approximately $4.1 billion),” and did so in just 20 minutes. While not named, the “mutual fund complex,” is Overland Park-based Waddell & Reed, rumored to be the primary culprit since the first weeks of the investigation. Waddell used an algorithm specifically designed not to drop the price on those E-Mini contracts (essentially futures on the movement of the S&P 500) by parceling out the huge trade so that it never rose above 9% of the total sales of E-Minis.
The problem, according to the report, was that the algorithm didn't take into account time, the price of the sales (outside of a lowest-price floor it wouldn't sell below) or tell the difference between volume created buy firms people trying to buy E-minis, which would make it easier for Waddell to sell them, and those trying to sell them, which would flood the market and make it harder for Waddell to sell.
In a simplified version of the tale, the initial buyers of all those contracts were mostly high-frequency traders, who essentially act as old-fashioned brokers on steroids—buying and selling massive numbers of stocks and futures, pocketing tiny trading fees, and getting the securities off their books as quickly as possible. So once they'd bought the contracts from Waddell, these traders tried to sell them. This drove up the overall volume of trading, which meant Waddell's algorithm could sell more contracts but still stay below that 9% threshold. The problem was that most market participants—from Waddell, to the firms who'd bought from Waddell in the first place—were trying to sell, which used up potential buyers. And since E-Minis are used to price all sort of assets, like exchange traded funds (ETFs) and the individual stocks that make up those funds, tremors in the E-Mini market turned into tsunamis in other markets.
When the report was first leaked on October 1, the press piled on to Waddell. “Flash Crash Is Pinned on One Trade,” read the Wall Street Journal headline. “Lone $4.1 Billion Sale Led to ‘Flash Crash’ in May,” said the New York Times.
But it wasn't just Waddell getting the heat. On NPR's Marketplace, Fortune magazine's Leigh Gallagher said, “there's just no suggestions for policy changes or regulatory framework that could stop this from happening again, which basically means it could happen tomorrow.” The popular and humorously hyperbolic finance blog Zero Hedge sarcastically concluded that: “Moral of the story: NO ONE MUST BE ALLOWED TO SELL MORE THAN ONE SHARE OF STOCK AT A TIME EVER!!! YOU WILL OVERLOAD THE MARKET, FLOOD THE NYSE'S LRP, CAUSE A LIQUIDITY CRISIS, DESTROY THE MARKET AND END CIVILIZATION AS WE KNOW IT.”
Since the Crash, Waddell's stock price has dropped more than 20%, and the company has been reduced to sending out BP-esque missives about how it didn't intentionally to blow up the world. But in hopes of clearing its name, Waddell provided Nanex with its own trading data from the Flash Crash (with Barclay's certifying the authenticity of the data).
With the new data in hand, Nanex has concluded that the regulators' report just doesn't fit the timeline: most of Waddell's selling took place after the Crash, during a rebound. “The algorithm was very well behaved,” reads a Nanex report released this afternoon. “It was careful not to impact the market...And when prices moved down sharply, it would stop completely.”
The Crash, says Nanex, was actually caused by the aggressive reselling of those contracts by other firms (Nanex doesn't say what kind of firms, but the SEC-CFTC report identifies them as high-frequency traders). “Rather than making sure the sale would not impact the market, they did quite the opposite: they slammed the market with 2,000 or more contracts as fast as they could...As time passed, the aggressiveness only increased, with these violent selling events occurring more often, until finally the e-Mini circuit breaker kicked in and paused trading for 5 seconds, ending the market slide.”
CFTC Commissioner Bart Chilton says that the reselling of the E-Mini's played a role in the crash, but he's confident in the report's finding that the Waddell trade started the downward spiral. “We looked at hundreds of thousands of trades, and this is the one that stands out. It was just one domino, but it was the domino that started the decline. It didn't fall any harder than any others, but it was first one to fall....Any stories that are trying to pin everything one one culprit like in a whodunnit mystery [aren't fair] but this trade was the start of things.”
And so the plot thickens, and the criticism is sure to continue thanks to the new findings from Nanex. Some of the piling on so far has been a bit unfair. While the report does single out the trade that allegedly sparked the crash, it doesn't scapegoat Waddell as much as news stories have reported. The vast majority of the report is actually a fairly readable breakdown of the mechanics of the Crash—a much more complex story that doesn't lend itself to headlines. And the report was never intended to be an omnibus prescription for preventing a similar collapse. The SEC-CFTC joint advisory committee is expected to meet some time later this month and offer recommendations that will be much more comprehensive.
“There are a lot of things we're going to look at,” said Commissioner Chilton. “Should there be limits on how much of a commodity you can trade and how quickly you can do it? That's not something that's going to be very popular with banks or hedge funds or much of anyone else, but there might be a level where trading should be capped,” to prevent the market from being flooded.
Still, the report has brought a lot of this criticism on itself. The recommendations it does make, and even many of the conclusions it comes to, seem incommensurate to the problems of the crash. The fact that the NYSE's computer systems couldn't keep up with trading volume and printed delayed and inaccurate stock quotes? A couple brief references buried in footnotes and deep in the report saying things like, “we do not believe significant market data delays were the primary factor in causing the events of May 6.” And no harsh words for the exchanges or demands that they fix the problems.
What to do about the lightning-fast reselling of futures that certainly contributed to (and in Nanex's analysis, caused) the Crash? It doesn't say much. What about the upside of high-frequency trading, firms who actually stayed in the jittery market and provided liquidity during the Crash, only to have exchanges like the NYSE cancel their trades and cost them millions? The report codifies a byzantine set of standards for canceling future trades, but they seem too complex for most trader to take into account during real-time trading, and are fairly moot, since an exchange has the right to cancel any trade it wants to. So in the unfortunate event of a repeat crash, many traders might be so afraid of having their trades canceled again that they'll simply pull out of the market entirely.
And finally there's the the bigger question of What This All Means. “Despite the knee-jerk reaction on part of anybody who wanted to get on TV,” says Illinois Institute of Technology professor Ben van Vliet, a high-frequency trading expert who works with the Chicago Mercantile Exchange and CFTC, the report shows “that automated systems had nothing to do with it....There's actually a much better argument that the reason the market came back so fast is automated trading systems. Automated systems, because they don't trade on emotion, calculated the probabilities, [bought the undervalued securities] and that's why things came back so quickly.”
In an important sense van Vliet is right: according to the report, it wasn't some rogue algorithm or computer gone amok that sparked the Crash, but the result of algorithms that traded the way their designers intended them to. But the report also lays bare a classic problem with automated trading: with all the myriad factors that go into a complicated algorithm, the law of unintended consequences can rear its ugly head, and given the speed of these programs, the damage has often been done before anyone even realizes what's happening.