Flash Crash
Page 15
The CFTC’s job was threefold: to substantiate Mr X’s findings; to establish what, if any, offences had been committed; and to build a case that would stand up in court. Marvine and the other attorneys could handle parts two and three. Part one was more challenging. Mr X’s information looked compelling, but it was incumbent upon the agency to independently verify the allegations, which was more difficult than it sounds. Simply re-creating the e-mini order book in the age of HFT, when billions of orders are placed, modified and cancelled each day, was a feat of computing power and human endeavour that had never been attempted on this scale by the agency before. Assuming it was possible, the team would then need to piece together the target’s trading strategy and demonstrate intent. The spectre of failed manipulation cases loomed large.
The Kansas City attorneys needed an investigator well versed in wrangling data, and when they heard Harris was available, they drafted her onto the team from afar. Since joining the CFTC in 2009, Harris had tried her best to keep up with the evolving markets. She’d worked on one of its first spoofing cases, meticulously piecing together wheat trader Eric Moncada’s manipulative trading on an Excel spreadsheet with ten million columns that slowed her PC to a crawl; and when the division invested in a new program called SAS (Statistical Analysis System) that allowed it to store, interrogate and visualise large quantities of data more effectively, she’d been among the first adopters. She enjoyed the field so much she’d signed up for an accelerated master’s programme in systems engineering at the University of Virginia, and she was now spending her nights and weekends taking pre-req classes in scientific programming, linear algebra and statistics.
‘To be honest, training at the CFTC was pretty minimal, and there weren’t a lot of people in the division who had experience in any kind of advanced statistics and who actually understood how to get data sources and analyse them,’ Harris says. Previously, Kirilenko and his fellow academics in the Office of the Chief Economist might have been able to offer some technical assistance, but the unit had just been decimated following the data breach complaint. ‘It was difficult to do the job without the right foundation. So I took it upon myself to get it.’
GROWING UP in Ann Arbor, Michigan, Harris always had a lot of questions. Her father, Charles, was a former Navy officer of Welsh descent from a long line of military men. He worked at the sprawling General Motors factory in Romulus, Michigan, and inspected the kids’ rooms before they went to bed. Her mother, Guerda, a nurse and, later, the owner of a catering business, was born in Haiti. As a child, Guerda had escaped François ‘Papa Doc’ Duvalier’s brutal regime under dramatic circumstances that were never openly discussed. The secrecy drove Jessica and her two siblings mad.
With their tract house, soccer practice and Friday night football, the Harrises blended in with the neighbourhood kids. ‘We dressed kind of preppy and it was hard to know where we were from,’ says her elder brother Chris. But on weekends, they were taken to get-togethers in Detroit and South Michigan where the adults drank rum and talked Port-au-Prince politics, and the smell of rice djon-djon and fried griot filled the air. Guerda came from a prominent family in Haiti before it was ripped apart by Duvalier – one of her brothers, Pierre François Benoit, would go on to become the country’s ambassador to the United States – and it was at these parties that the children first caught snippets about their mother’s life on the Caribbean island. ‘There were always unanswered questions, a lot of mystery,’ says Chris.
The children knew better than to pry, but Jess was a natural investigator. When she was twelve, a stranger arrived at the family home in a BMW claiming to be one of Charles’s cousins. ‘Jess just looked him up and down and pegged straightaway he was a mooch,’ says Chris. To the kids’ irritation, the distant relative hung around for weeks, eating and sleeping for free, talking up his latest plans and inviting himself to gatherings. ‘The only thing he cared about was his car,’ says Chris. ‘Then one day he comes in shouting because someone had snapped the antenna.’ Jess was tight-lipped about the incident, but a few days later the interloper left, defeated.
Harris studied business administration at Western Michigan University and then landed a job as an assistant to a commodities broker at MF Global in Chicago. It was her first exposure to the world of trading and, while she got a kick out of the pace of the markets, she found the work soulless. So when she read about an opening at the National Futures Association, she decided to apply. The NFA is a quasi-regulatory body that oversees futures brokerages and trading firms, investigating misdemeanours that either aren’t serious enough for or haven’t yet been referred to the CFTC. Harris felt instantly at home there and rose rapidly from auditor to investigator, then to senior investigator and then to supervisor. She walked into firms with a name badge, sat down with company directors twice her age and asked them whatever she wanted. ‘She wasn’t just good. She was stellar,’ says Dan Driscoll, who is now chief operating officer of the NFA. ‘Really, one of the most determined people I’ve ever met. A complete bulldog.’ In 2009, Harris was working on a fraud case that involved liaising with the CFTC, and when the case wrapped up, a manager at the agency asked her if she’d be interested in joining. ‘It’s a real Cinderella story, I know, right?’ she laughs.
HARRIS’S OFFICE was sparse and functional: a couple of whiteboards scrawled with charts and formulae, some big stacks of paperwork, a bike helmet hanging on the door handle. Outside the floor was buzzy, but Harris spent most of her days alone, staring at her screens, immersed in data. Every once in a while, a colleague popped their head around the door to ask a question or say hello. ‘Truth be told, I might as well have been in my mom’s basement drinking Mountain Dew in my sweatpants or something,’ she says. After landing the Sarao case, Harris’s first task was to pull in trading data from various sources. Historically, the CFTC’s first port of call on manipulation cases was something called the ‘Trade Capture Report’, or TCR, which listed every completed or partially filled trade. Kirilenko and his colleagues had relied on TCR to identify Waddell & Reed’s errant algo in the days after the Flash Crash, and had used it as the foundation of their joint report with the SEC. But modern trading was as much about cancelled orders as consummated ones, and the patterns Mr X had identified – the rapid-fire placing and modifying of blocks of orders several ticks away from the prevailing price – didn’t show up since, by design, the orders were almost never hit. To isolate the spoof orders, Harris had to use two additional data sets provided by the CME called Rapid and Armada. Rapid logged every order placed by an entity, regardless of whether it was filled, partially filled, modified or cancelled, allowing Harris and her colleagues to view their target’s activity in all its complexity. Armada enabled her to rebuild the ladder, nine levels above and nine below the best bid, and offer at any given moment, helping her place the trader’s actions in the context of what was going on around them.
Rather than identifying traders by name, the CME allocated each entity an ID code, and for the first few weeks of the investigation, the target was known only as NAVSAR. Based on the size of the orders and the scope of the activity, the team speculated as to whether they were dealing with a large HFT firm or a big-shot trader at a Wall Street bank. They could see the participant traded through MF Global, and under normal circumstances they would have been able to get the entity’s details from the broker quickly. However, since MF Global was in bankruptcy proceedings, all they could do was write to the liquidator and wait. In the meantime, Harris continued to work the data.
NAVSAR’s activity in the e-mini was consistent with spoofing, but many trading strategies involve placing bets across multiple markets, and it was entirely possible the e-mini trades were part of some cross-asset approach she wasn’t aware of. Harris ruled that out quickly when she ascertained that her subject was usually active only in the e-mini. Next, she pulled together a series of distributions to determine how the entity’s trading compared with the rest of the market. ‘You want to fi
nd out what’s normal,’ she explains. ‘What’s the average trade size in the e-mini? What proportion of orders is cancelled? How long do orders usually rest in the order book? What is an outlier?’ NAVSAR may have been trying to manipulate the market, but if everyone else was doing the same thing or worse, it would be hard to justify bringing a case.
By any measure, NAVSAR was an outlier. In the twelve days the CFTC ended up selecting to illustrate the entity’s activity, its layering algorithm cancelled or modified orders 182,000 times, corresponding to $35 trillion in notional trades – double the size of America’s gross domestic product. On eight of those days, not a single one of those orders was hit. The size of the orders was also immense: an average of 504 contracts, where the average across the market was seven. On the day of the crash, the layering algo accounted for close to a third of all cancelled trades in the e-mini, the second-biggest futures market in the world. ‘Undoubtedly, we were looking at the biggest spoofer that has ever traded,’ says Mr X.
In March 2013, the account documents came back from MF Global’s trustee. They showed that NAVSAR represented an entity called Nav Sarao Futures Limited, a British company with a single director. Nobody at the agency had heard of the firm, but new prop shops sprouted up all the time. It was only when the team looked up the address on Google Maps that eyebrows were raised. Did one of the five biggest S&P traders in the world really operate from a semidetached house in suburbia? While Harris was busy with the data, the attorneys in Chicago started interviewing Sarao’s brokers and others from across the market. Kernels about the trader’s eccentricities came in slowly: how demanding he was, his abstemiousness. One broker relayed how Sarao had got lost on the way to their offices and needed to be guided in, step-by-step, on the phone. Others painted a more sinister picture. A straitlaced surveillance employee at the CME told the story of how Sarao threatened to fly to the United States and ‘cut your fucking thumbs off’ if he didn’t sort out an issue he was having. Harris and her colleagues couldn’t tell if they were dealing with Forrest Gump or a character from a Guy Ritchie movie.
As work on the case progressed, the team encountered another roadblock, this time internal. Ultimate responsibility for overseeing the case resided with a manager who had been with the division since trading was still carried out in the pits. He had never been convinced of the case’s merits, and in tense meetings, he pooh-poohed the idea that Sarao could have had such an impact on the market. The debate was ostensibly focused on the technicalities of proving that injecting so-called false liquidity actually moved markets; but all conversations around Sarao had an unspoken subtext: How would it look if the CFTC turned around and suggested that spoofing had helped cause the Flash Crash, when three years earlier it published a report that didn’t even mention it as a factor?
By autumn 2013, the team had verified and built on much of Mr X’s evidence, identified Sarao and pulled together some of his interactions, including the warnings he’d received from the CME about his trading during the pre-market. It was a promising haul, but they knew it wasn’t sufficient to nail him. Spoofing was a brand-new offence, completely untested in court. And meeting the standard for manipulation required them to prove that Sarao intended to cheat, an ominous cliff face to climb. There were also aspects of the case that gave them pause. If what they were seeing was as clear-cut and persistent as it appeared, why hadn’t the surveillance people at the CME taken any substantive action? And how had MF Global allowed a day trader in his bedroom to trade with such insanely high stakes?
Unlike the criminal authorities, the CFTC had no power to subpoena or exercise a search warrant on Sarao, a non-US citizen. Faced with an impasse, they made a decision to reach across the void and make contact with their quarry. Harris and one of the attorneys drafted a letter containing open-ended questions designed to yield information without alerting Sarao to the existence of a full-blown investigation: tell us about your trading career; what software do you use; what is your strategy. Beneath the gentle language was an unmistakable message: We’re on to you.
CHAPTER 19
CORNBREAD AND THE CME
Nav’s first instinct upon receiving the CFTC’s letter was to ignore it. Every once in a while he got a missive from one of the exchanges asking him about his trading practices, but nothing ever came of it, and he was inclined to view any contact from the regulators as worthy of disdain. A few weeks earlier, R. J. O’Brien had forwarded him the CFTC’s ‘Interpretive Guidance … on Disruptive Trading’ document, which detailed exactly what was and wasn’t allowed under the spoofing rules. Rather than reflecting on his own tactics, Nav wrote back to his broker: ‘Lol, guarantee if I switch on my computer I’ll see the same people breaking all those rules, day in, day out.’ When he told MacKinnon and Dupont about the letter, though, they implored him to take the matter seriously, and in December 2013, MacKinnon took Nav to meet three lawyers in London, who all told him the same thing: ignore this at your peril. Nav nodded along when they offered to engage with the CFTC on his behalf, but after discussing fees he opted not to take them on. Shortly afterwards, a group of traders in Chicago filed a lawsuit against the CME Group that resonated deeply with Nav and played a part in how he chose to respond to the authorities’ inquiries.
AT THE Chicago Board of Trade, where it was an unwritten law that everyone had to have a nickname, they used to call Bill Braman ‘White Cornbread’, then just Cornbread, a riff on his polite ‘aw, shucks’ demeanour and the WCB initials on his name badge. Braman, who is tall and lean with a shaved head, joined the Treasury pit from college just before Black Monday in 1987, and stayed there until the summer before 9/11, when he started trading on a screen. The transition was tough, and many of his friends fell by the wayside, but Braman was a quick, determined learner and before long he was earning more money than he ever had. ‘I had the best years of my life because I knew how to trade and the banks and the smart people hadn’t figured out the electronic side of it yet,’ he recalls. ‘I loved to go to work every morning at 5 a.m. and be done by lunchtime, have a workout, and come home to pick up the kids from school. It was a nice life.’
Sometime around 2007, things started to change. As high-frequency trading proliferated, the strategy Braman had honed, which involved betting on relative movements between bonds with different time horizons, became less and less effective. He tried to adapt, incorporating algorithms and investing in technology that allowed him to execute trades more quickly. He even taught himself some code. But the HFTs were too fast, gobbling up opportunities before anyone else could react. Eventually, in 2012, Braman gave up and left the industry that had come to define him. Within a few years, he was divorced and working as a barista at Starbucks. In his spare time, he developed a bicycle dog leash called the DoggerJogger. ‘Going from millionaire to broke doesn’t help your family life very much,’ he says. ‘I know people who have killed themselves. They go from driving fancy cars and owning big houses, and they can’t face their family and friends.’
Braman says there are thousands of dispossessed traders like him, the diaspora from the once-heaving amphitheatres of the Chicago Board of Trade, LIFFE and the CME. He is philosophical about the evolution to screen-based trading, viewing it as part of the march towards mechanisation that touches all industries. But the ascendancy of HFT has been harder to take. ‘Imagine we’re in the pits and I have a really good spot next to the broker and I can hear all the trades coming in, and then one day two six-foot-eight guys come along and stand in front of me and crowd me out and I can’t see the trades like I used to,’ says Braman. ‘It used to be a meritocracy. If you traded big and took big trades and could service the brokers, they’d make room for you. HFTs aren’t big guys. They’re just weasels who were able to sneak in, listen to what’s about to happen and jump in front of it while taking no risk themselves.’
Worse than the HFT firms, to Braman’s mind, is the CME Group, the $75-billion corporation that owns the formerly member-owned exchanges and w
hose directors got rich by letting in the vampires to bleed them all dry. So when an old pit buddy known as ‘Missile’ called and asked if he was interested in joining a class-action suit against the CME, Braman was receptive. The suit was the brainchild of a former pit inhabitant named R. Tamara de Silva who had parked her own dreams of making it as a trader in order to train as a lawyer. De Silva believed that, by giving preferential treatment to select customers, the CME Group was perpetrating fraud on a massive scale. But convincing her buddies from the market to sign up as plaintiffs wasn’t easy. ‘You’ve got to understand, Chicago is a trading town and the Merc is very powerful,’ says Braman. ‘People are genuinely afraid of it, like the mafia or something.’ Cornbread figured he had nothing to lose and agreed to add his name to the complaint.
De Silva had been admitted to the Northern Illinois federal trial bar only fairly recently, in 2007, and she had limited experience with cases relating to the Commodities Exchange Act. What she did possess was a sense of righteous self-confidence that pervaded all aspects of her life. On her website, which featured a logo of her name with the scales of justice jutting out of the R, de Silva described herself as ‘an aggressive and fearless advocate’ who would ‘not take no for an answer’ and who specialised in bringing ‘cases that other lawyers will not take’. She also kept a blog, ‘Timely Objections’, containing heated takes on the issues of the day. Outside work she ran a kennel called Mythmaker that bred spaniels whose exploits she chronicled on Instagram. Braman likened taking on the world’s biggest commodity’s exchange to the movie Erin Brockovich and, with her shock of golden curls, de Silva made a fitting lead.