Flash Crash

by Liam Vaughan

  The first iteration of Braman et al. v. The CME Group Inc. was filed on 11 April 2014. In it, de Silva alleged that the CME had entered into ‘clandestine contracts’ with HFT firms, allowing them ‘to see price data and unexecuted order information before anyone else in the financial world’ while publicly claiming everyone got it in ‘real time’. In doing so, she wrote, the CME was guilty of manipulation, fraud and providing false information. The filing was a sparse seventeen pages and offered no evidence to substantiate its claims. It also contained some questionable assertions, such as that HFTs generally entered very large orders: in reality, most traded little and very often.

  Two days later, the CME issued a withering response: ‘The suit is devoid of any facts supporting the allegations and, even worse, demonstrates a fundamental misunderstanding of how our markets operate. It is sad when plaintiffs’ lawyers bring a suit based on a desire for publicity, and in the rush to file a suit fail to undertake even the most basic effort to determine if there is a basis for their allegations.’

  One thing de Silva’s suit did have going for it was timing. The previous month, Michael Lewis, celebrated author of The Big Short and Moneyball, had published Flash Boys: A Wall Street Revolt, a devastating exposé of the stock market and high-frequency trading. Through the character of Brad Katsuyama, an archetypal outsider determined to change the system, Lewis described how markets had devolved into a monstrously complex and fragmented realm in which speed was everything and legalised front-running was the norm. ‘The US stock market was now a class system of haves and have-nots, only what was had was not money but speed (which led to money),’ he wrote. ‘The haves paid for nanoseconds; the have-nots had no idea that a nanosecond had value. The haves enjoyed a perfect view of the market; the have-nots never saw the market at all.’

  Lewis wasn’t the first to question the fairness or utility of modern markets, but he had an audience, and by stating the case so starkly, he brought the debate about HFT – a debate that had flickered and then fizzled out in the aftermath of the Flash Crash – back into the mainstream with a bang. Flash Boys was an instant bestseller, and within a week the DOJ and the SEC both announced they were opening investigations into the HFT arena. Launching his own probe, New York attorney general Eric Schneiderman dubbed HFT ‘insider trading 2.0’. The exchanges and trading firms pushed back, decrying Lewis’s book as sensationalist and willfully misleading, but many financial insiders sided with the author. Joseph Stiglitz, the Nobel prize-winning economist, described HFT as a form of rent-stealing that distorted markets and discouraged investment. Charlie Munger, Warren Buffett’s right-hand man, said it was ‘the functional equivalent of letting rats into a granary’.

  Arriving in the middle of this groundswell, de Silva’s suit against the CME attracted widespread media attention. However, futures markets and stock markets are built on fundamentally different architecture, and, as the CME and others were quick to point out, many of the structural issues laid bare in Lewis’s book simply didn’t apply. For one thing, while the shares of major US companies could be bought and sold on dozens of different venues, most futures, including the e-mini, were traded on a single exchange. As a result, many of the scenarios Lewis described, involving predatory algorithms waiting for an order to hit one exchange, then racing over to the next, could never happen in futures.

  Still, as loudly as CME chief Terry Duffy insisted his markets were immune to the kind of systematic front-running that beset equities, the fact remained that between 2006 and 2013 high-frequency trading had gone from representing 30 per cent of US futures market volume to more than 60 per cent. It begged the question: if HFT didn’t have an edge, then what were they doing there?

  Under something called the Federal Rule of Civil Procedure 9(b), plaintiffs alleging fraud in the United States are required to lay out the basis of their case ‘with particularity’, a higher bar than for other offences, and in May the CME filed a motion to dismiss the case for essentially being too vague. Before the judge could rule, de Silva asked for a chance to refile and started contacting other attorneys who might help strengthen her claim. Among them was Lovell Stewart, a firm founded by a pioneering commodities lawyer named Chris Lovell, who had built a practice helping parties affected by manipulation and price-fixing bring civil claims for damages. Lovell and his colleagues had successfully sued a number of exchanges and, while they perceived de Silva’s filing as flimsy, they could see the merits in suing the CME. After speaking to the plaintiffs and carrying out their own research, they agreed to help redraft the complaint.

  Cornbread, Missile and their peers contended that HFT firms benefited from a host of disclosed and undisclosed advantages gifted to them by the CME Group, which had an interest in keeping volumes as high as possible. The problem the plaintiffs had was that any data that might help illuminate the issue was closely guarded by the CME itself. If they could convince a judge that there was a case to answer, they could force the exchange to hand over evidence as part of a pretrial discovery process. Their best chance, the lawyers at Lovell Stewart advised, would be to consolidate everything already in the public domain and mould it into something compelling.

  The bedrock of the amended complaint was a May 2013 article in the Wall Street Journal titled ‘High-Speed Traders Exploit Loophole’, which revealed how firms that paid to connect directly to the CME’s servers were able to profit by receiving confirmation of their own trades a few milliseconds before it reached the so-called ‘public tape’ and using it to ascertain which way the market was about to move. ‘If crude oil is selling for $90 on the CME, a firm might post an order to sell one contract for $90.03 and a buy order for $89.97’, the Journal wrote, citing one of several potential uses of the loophole. ‘If the sell order suddenly hits, the firm’s computers detect that oil prices have swung higher. Those computers can instantly buy more of the same contract before other traders are even aware of the first move.’ This ‘latency loophole’ was an open secret within the HFT industry, the Journal reported, with Jump Trading and DRW Trading both acknowledging they utilised it, and ‘a person familiar with the thinking’ of Virtu Financial suggesting it was actually good for the market because it increased liquidity. It certainly seemed to be good for Virtu, which disclosed in its 2014 IPO prospectus that it had just one losing day out of 1,238 in the five years to February 2014.

  CME Group promised to try to minimise the lag and pointed out that anyone could pay the $75,000 a month it charged for direct access to its servers, but the plaintiffs insisted the HFTs’ advantages went beyond just speed. Citing confidential witnesses, they alleged that the CME had entered into ‘clandestine’ fee deals with select firms that slashed their cost of trading relative to the majority of participants, resulting in a ‘two-tiered marketplace’. The CME published details of its clearing fees on its website, but it also negotiated private ‘incentive’ programmes that were highly valuable and, according to the complaint, susceptible to conflicts of interest, given that some of the exchange’s biggest customers were firms operated by its own board members. The CME maintained that the deals were essential to bringing liquidity to the markets, but, as a Bloomberg article cited in the complaint pointed out, they were in place in some of the busiest markets in the world. It was akin to suggesting tourists needed to be paid to visit Times Square.

  Last, Braman et al. alleged that the CME Group was turning a blind eye to manipulative and disruptive trading practices that were ‘integral’ to HFT, such as wash trading and spoofing. Wash trading is when an entity buys or sells securities from itself to push the market around or hit volume targets so it qualifies for rebates, and by 2013 it had become widespread enough that the CFTC began an investigation. Publicly, the exchanges supported the clampdown, but they made so much money from all the additional buying and selling that they had little motivation to enforce it. And since most HFT firms had dozens of accounts, there was nothing to prevent them from transacting with each other and then claiming they we
re running unrelated strategies. With regard to spoofing, there was little evidence, despite the change in the law, that the practice was on the wane. By now, an estimated ninety-five out of every one hundred e-mini orders was cancelled, creating, in the plaintiffs’ words, a ‘false impression to other traders about supply and demand’. Sanctions remained rare, and even serial offenders escaped with fines in the tens of thousands of dollars.

  The revised complaint was lodged on 22 July, and it significantly upped the stakes. This time the plaintiffs claimed to represent a class that included anyone who had ever bought or sold a future or paid for data from the CME between 2005 and 2014, a group that included virtually every financial institution, farmer and food producer in America; and they demanded not only restitution for everything they’d ever lost, but the effective dismantling of the entire HFT industry. The opening paragraph was a call to arms:

  Over the past decade, the Chicago derivatives markets have engaged in agreements with certain high-frequency trading firms to erode the integrity of the marketplace and manipulate prices. These exchanges, together with a sophisticated class of technology-driven entities known commonly as ‘high-frequency traders’ have provided and utilised information asymmetry along with clandestine incentive agreements and illegal trading practices to create a two-tiered marketplace that disadvantages the American public and all other futures marketplace participants, all the while continuing to represent to the public and their regulators that they continue to provide transparent and fair trading markets to the global market. In reality, the advantages given to HFTs by the Exchange Defendants effectively create a ‘zero sum’ trading scenario where the HFTs gain what the Class Members lose by effectively providing HFTs with the opportunity to skim an improper profit on every futures transaction.

  FOUR THOUSAND miles away, Nav read the complaint and was filled with a sense of vindication. There, over fifty-nine pages, was everything he’d suspected but never been able to prove: the secret deals, the front-running of orders, the race for speed, the industrial-scale manipulation. The CFTC had recently sent him another letter, this time via the UK regulator, asking if he’d be willing to come in for a voluntary interview. But how could anyone criticise him when the whole show was a racket from top to bottom? Over the years, Nav had made videos of the markets in an effort to capture evidence of cheating by his competitors. Now he cranked up the recording, filling hard drives with footage accompanied by an expletive-strewn running commentary. He talked to his advisers about sending the material to the plaintiffs, and even considered joining the class action himself. In the end, he decided it would be wiser to keep a low profile. But the criticism surrounding the HFT industry gave him a renewed sense of confidence about his own situation, and on 29 May, in lieu of an interview, he finally emailed the regulators back.

  NAV WROTE:

  Below is the answers to your questions as well as a general overview of my trading and how I would advise the market could become more fair and efficient.

  I am an old-school point and click prop trader. To this day I am still using the mouse to trade. That is how I trade, that is how I always have traded, admittedly very very fast because I have always been good with reflexes and doing things quick … I am a trader who changes his mind very very quickly, one second I am prepared to buy the limit of 2,000, the next second I may change my mind and get out … This is what is unique about my trading … What makes me change my mind? Well it could be anything, a move in one of the other markets that I look at, a chart set up that I suddenly remember from my 11 years of trading, or simply the WAY I was filled made me doubt my position, or for the large part it is just my INTUITION.

  Nav went on to describe himself as

  a guy … who is trading from the UK and whose system is miles too slow compared [to the HFTs, and whose] orders have to travel further than everyone else’s who are trading in USA. No wonder they can manipulative [sic] on top of my orders without any risk … I don’t like the HFT arena and have complained to the exchange numerous times about their manipulative practices, please BAN IT.

  Asked about his setup, Nav wrote:

  I have traded using a basic TT for numerous years. Due to the fact that there were some individuals in the emini SP who quite remarkably seemed to know WHERE 100% OF MY ORDERS WERE RESTING … I decided to pay Edge Financial to build a program for me that would help disguise my orders more effectively … I do not know if this can be described as HFT, to me it is just giving me the ability to have some extra functions … It is called NAVtrader, but it could be called anything and I was the only one who helped design it, albeit my design ideas were 100% generated from what I had already seen other traders using …

  Nav went on to describe the JOIN, ICE and SNAP order types, the last of which he said had

  worked rather beautifully when the mass manipulator of the e-mini sp was doing his normal manipulative activity at price 1800.00 on Friday 24th January circa 12.23pm.

  Describing the layering algo, the tool that placed big blocks of orders several ticks from the prevailing price, Nav wrote:

  This is to catch any blips up/down in the market so that I can make a small profit as the market comes back into line (almost immediately). These orders are placed rarely and only when I believe the market is excessively weak or strong.

  All told, it was a fairly accurate description of Nav’s trading setup, albeit with some serious omissions. There was no mention of NAVTrader’s all-important ‘back of the book’ feature, which pushed orders to the back of the queue every time another order arrived; and his suggestion that he occasionally layered the ladder with orders several prices from the best bid to catch ‘blips’ was hard to square with the reality that he used the layering algo most days, and the orders it posted were consummated less than 1 per cent of the time. The ‘mass manipulator’ he referenced was Igor Oystacher: even when trying to placate the authorities, Nav couldn’t resist crowing about getting one over on his rival. The overriding message was that any dubious-looking trading on his part was really just a reflection of his indecision and lightning-fast reactions. Oblivious to the forces mounting across the Atlantic, he optimistically signed off:

  I hope this is helpful and sufficient for you.

  All the best,

  Nav.

  CHAPTER 20

  MINDGAMES

  The clock was ticking for Jessica Harris and her colleagues at the CFTC. More than four years had passed since the Flash Crash, and the agency was subject to a five-year statute of limitations, meaning if they were going to charge Sarao for his actions that day, they needed to act fast. Sarao had continued to trade, of course, but it was his activities on 6 May 2010 that caught the attention of Mr X, and the agency was keenly aware it was Sarao’s association with the crash that elevated just another manipulation case into a blockbuster. The team had pulled together everything they could, but the reality was that the evidence still looked light. Beyond the trading records Harris had prepared, there were some dubious messages between Sarao and his brokers, as well as a handful of testy exchanges with the CME on the need for bona fide orders. They painted a compelling picture, but nobody felt confident it would be enough to meet the ‘by a preponderance of the evidence’ standard applicable in civil cases. When Chuck Marvine, Harris’s boss, was promoted in spring 2014, he was relieved to hand responsibility for the case to a senior prosecutor in Kansas City named Jeff Le Riche.

  Le Riche had grown up obsessed with science and space in a small town in rural Missouri, devouring Carl Sagan and Stephen Hawking books. At university, he majored in chemistry, but he came to the conclusion he wasn’t going to cut it as a scientist, so he went to law school instead. He worked in private practice for a while, then, after his kids were born, joined the CFTC so he could spend more time at home. When electronic trading exploded and markets grew bigger and more complex, his grounding in statistical analysis and data came to the fore and, like Harris, he found himself in demand.

  One o
f the CFTC’s abiding concerns was that Sarao’s methods, and the high-frequency trading arena in general, were too complex for a layperson to comprehend. It was easy to imagine a jury being unwilling to condemn someone for a crime they didn’t fully understand. To help cut through the noise, Le Riche hired a University of California, Berkeley finance professor named Terry Hendershott as an expert witness. Hendershott’s main job was to analyse the most blunt and readily understandable weapon in Sarao’s arsenal – the layering algorithm he’d developed with Hadj at Trading Technologies. The professor examined Sarao’s use of the algo over a sample of twelve days between 2010 and 2014 and concluded it was consistent with spoofing. The orders the algo placed were comparatively huge, rarely if ever hit, and were likely to have had a ‘statistically significant’ impact on prices – around 0.3 basis points on a thousand-lot order. And while Hendershott noted there were legitimate reasons to layer an order book, such as market making or hedging another position, he said Sarao’s trading bore none of their characteristics. ‘Based on the above analyses and my experience the Layering Algorithm’s massive volume of orders were designed not to execute and there is no clear business motivation for wanting those orders to execute,’ he concluded in a report. On the subject of the Flash Crash, he was more circumspect. Resisting pressure from the CFTC to tie the day’s events to Sarao, Hendershott would only note, somewhat tortuously, that ‘the layering algorithm contributed to the overall Order Book imbalances and market conditions that the regulators say led to the liquidity deterioration prior to the Flash Crash’.