So now there’s a new theory about the “flash crash” on 2010. I thought we had this one straightened out. It has passed out of the realm of journalists, into that of historians. It was Waddell & Reed. Right?
What puts it very much into the former realm, though, is the arrest of Navinder Singh Sarao in London, and charges by the Commodity Futures Trading Commission against both Sarao individually and his firm, Nav Sarao Futures Ltd.
The CFTC complaint, filed with the U.S. District Court for the Northern District of Illinois on April 17th, came unsealed on the 21st. It says that the defendants unlawfully manipulated the E-mini S&P 500 near month futures contract. How? By the use of fancy algorithms. This involves a variant of “spoofing” called “layering.”
The New Theory
The allegedly illegal behavior involved what the CFTC calls a “commonly used off-the-shelf trading platform” that the defendants modified to layer from four to six very large “sell” orders into the visible E-mini S&P central order book. Each sell order was one price level from the other. As the price moved, the algorithm modified the sell orders too, so they didn’t become the best asking price but did remain visible to other traders. Eventually, the orders would be cancelled without any resultant transactions.
The idea was to overload the sell side of the order book and artificially lower the E-mini market price. “When Defendants turned off the Layering Algorithm, the artificially lowered price typically rebounded to its previous level.” Defendants could make a profit in both directions.
The complaint identified 12 “Example Price-Impact Days,” which it uses as paradigms/illustrations of defendant misbehavior, while adding in a footnote that this misbehavior “was not limited solely” to those days. We’re told that on some of the EP-IDs, “when the Layering Algorithm was active, Defendants’ sell-side orders constituted as much as 40% of all active sell-side orders.”
Further, the shenanigans that set off the flash crash didn’t end then. The “layering” allegedly continued until as recently as April 6, 2015. The last of the EP-IDs is March 10, 2014.
Still, the day that has lived in market-structure infamy was another of these dozen examples, May 6th, 2010. It was at 2:42 that day that the Dow Jones began a stunningly rapid fall, dropping 600 points in five minutes. Just twenty minutes later, that is by 3:07 that afternoon, it had recovered most of the 600 points.
Remember Greece?
The year 2010 was one much like 2015 in certain particulars. For example, the financial world spent a lot of its mental energy watching and worrying about Greece that year. In March 2010 the government of Greece announced a package of austerity measures designed to demonstrate the country’s credit worthiness in the face if investor skepticism. The package didn’t work. In April, S&P downgraded Greece’s credit rating below investment grade.
In the words of one scholar at the Athens University of Economics and Business, this was the moment when Greece “effectively lost access to the international financial markets” so that “a sovereign debt crisis threatened to develop into a solvency crisis.”
Markets were jittery about this, and the Dow was already down significantly on May 6, 2010, in the hours before the 2:42 drop off. [The above chart of the Dow’s Dow movements comes from a presentation of Blair Hull at the Chicago Trading Show 2012.] This was a likely day for something dramatic.
The Old Theory
But as to the proximate cause of the flash crash, attention has focused for a long time now on a big sale by Waddell & Reed. The official report issued by the SEC and the CFTC on September 3, 2010 referred to W&R as a “Large Fundamental Seller” that was using an algorithm of its own. This algorithm was a simple one, selling into large volumes. It responded to increased volume in E-minis by selling more of W&R’s own into the market, setting in motion the cascade leading to the falling-off-a-cliff effect one observes in the chart above.
Does the CFTC’s new theory mean it has let W&R off the hook?
Well … yes and no. And yes. First, yes, the new account is sufficiently different from the old one that it confirms at the least some of what skeptics about the old account have said all along. But second, and on the other hand, events like this need not have a single unique cause. W&R, as well as Greek-bond related jitters, Sarao’s layering, etc., might all have been factors contributing to a given outcome.
But third, yes. Let’s call W&R “off the hook.” The case against them was already looking threadbare, and by now is positively moth eaten. Instead of looking for particular bad guys, perhaps the inquiry ought all along to have been focused on issues of market structure. And if we believe that, then switching from one scapegoat to another isn’t an example of progress.