Flash Crash

by Liam Vaughan

  Part of the problem was cultural. Since futures markets, unlike stocks, were almost exclusively inhabited by professionals, an ethos of caveat emptor prevailed. Regulation in this clubby world was considered unseemly, and the exchanges, at least nominally, took the lead in keeping everyone in check. Every now and then, after a spike in oil or a collapse in wheat prices, the political pressure to intervene became too much and the CFTC announced a probe. But most of the cases it brought were decidedly small fry. From its inception, the agency was timid, underresourced and, crucially, hamstrung by the law.

  When Congress passed the Commodity Exchange Act in 1936, it banned manipulation but declined to define what manipulation was. That task was left to the courts, which came up with different, often contradictory interpretations. How much and for how long did prices have to move to constitute manipulation? Was it necessary to show a participant meant to impact prices, or would recklessness suffice? What distinguished cheating from legitimate trading based on supply and demand? In a landmark 1982 case involving an entity called the Indiana Farm Bureau, a four-part test was formulated that laid down a heavy precedent. For manipulation to occur, the court said, prosecutors had to prove that a party had the ability to influence prices; that they specifically intended to move them; that artificial prices existed; and that the defendant caused them. This turned out to be an almost impossible threshold to meet, and over the next four decades, the CFTC prevailed in just one manipulation case at trial. Proving a price was artificial and determining what caused it was an imprecise, knotty endeavour involving price regressions and statistical analysis that was hard to explain to a jury and easy to refute. The agency had more success prosecuting the lesser charge of ‘attempted manipulation’, which rested on the first two of the four legs, but even then, when it came to intent, traders could usually find some reasonable-sounding explanation for their actions. As a result, the majority of cases either settled early with a modest fine or collapsed before making it to trial. After one particularly high-profile failure, a group of senators wrote to the agency to say it was ‘an embarrassment’. ‘The whole purpose of the CFTC was to stop manipulation and it was toothless,’ says legal scholar Jerry Markham, the author of a paper that characterised commodity manipulation as ‘The Unprosecutable Crime’.

  When Gensler was appointed chairman in 2009 following a drawn-out and contentious nomination process, the former banker pledged to plug the gaps in the manipulation rules and bring the CFTC’s regime more in line with the SEC. Members of the Enforcement Division came up with something called ‘Disruptive Trading Practices’, which identified acts they saw as antithetical to the fair functioning of markets. These included what’s colloquially known as ‘banging the close’, that is, trading heavily during the short window when the settlement price of a commodity is set to benefit another position; and spoofing, which they defined as ‘bidding or offering with the intent to cancel the bid or offer before execution’. By criminalising specific behaviours, the CFTC hoped to provide its prosecutors with an alternative route to victory at trial, and with little fanfare an amendment was inserted onto page 684 of the draft Dodd-Frank bill in December 2009. The existing manipulation statutes in the Commodity Exchange Act were also tweaked to make it easier for the government to prevail.

  At first, few people in the industry seemed to pay much attention, but around a year later, at a roundtable discussion at the CFTC’s offices in Washington, several parties, including the exchanges, raised objections. They argued there was no widely understood practice called spoofing in futures and that the CFTC’s definition was too vague. They also said that instituting a blanket ban on placing orders you didn’t want to consummate risked inadvertently prohibiting legitimate practices like placing stop-losses.

  It’s part of the dance of rulemaking that industry participants tend to object to new rules regardless of how sensible they might be, but the HFT community got behind the spoofing ban straightaway. Adam Nunes from the New York firm Hudson River Trading said the proposal would ‘make the market more liquid and more efficient’, and ‘allow legitimate practices to occur without the risk of being manipulated’. This, he said, would be ‘good for end users and good for, you know, firms like those around the table’. Nunes’s response makes sense when you consider that spoofing in the electronic era specifically targeted firms like his – HFTs that parsed the order book for information that would allow them to predict which way the market was about to move and then traded ahead of it. Their system only worked, though, if the signals the robots were scanning were an accurate representation of their opponents’ intentions. If traders disguised their plans by placing sell orders in the ladder when they planned to buy or, like Nav, inputting mammoth orders several ticks from the current price that they had no real desire to consummate, the machines were thrown off-kilter. In a fair and open market, the HFT firms argued, participants had a right to believe that the orders they were seeing were genuine. Not everyone agreed.

  ‘The way regulators think is fundamentally different to the way traders think,’ says Albert Kyle from the University of Maryland. ‘Traders think of trading as a game. Many traders have a gaming background: poker, backgammon, computer games. You make moves. I’ll bluff and I’ll try and maneuver around to get my strategy implemented. Regulators were suddenly talking about it as though it was a bunch of gentlemen trying to accurately communicate what their intentions are.’

  Markham, who, like Kyle, had worked at the CFTC before entering academia, believes there is a crucial difference between deception, which has been part of trading since markets began, and manipulation, which involves deliberately pushing prices around. ‘This is not a moral issue,’ he wrote in a legal filing opposing the spoofing rules. ‘Trading is a competition and … concealment of actual trading strategies is an integral part of that … In [American] football, concealment of the actual strategy for each play is critical to success,’ ergo Statue of Liberty plays and quarterback sneaks. ‘In volleyball, the setter tries to fool opponents on where the ball will be placed for return. Baseball pitchers disguise their pitches to fool batters. Hockey players try to deceive the goalie as to where the puck will be sent, and on and on. Trading in financial markets is no less a competition.’

  What’s more, says Markham, it’s hard to see how an order can be deemed manipulative or ‘non bona fide’ in a world where algos are able to react to fresh information in a matter of microseconds. One option the CFTC briefly considered was to remove any subjectivity from the process and simply limit the proportion of a participant’s orders they’re allowed to cancel. Anyone who breached the threshold would receive a fine, or worse. Another idea was to mandate that all orders have to remain in the order book for a minimum of, say, three seconds before they can be pulled. Ultimately, though, these approaches were deemed too prescriptive and were discarded in favour of what’s known in the legal profession as ‘scienter’, that is, what’s going through a trader’s mind when he places an order. It was a controversial step. ‘Prior to this there was no law that said you’re not allowed to think certain things while you trade,’ says Michael Kim, a white-collar defence attorney and former Justice Department prosecutor whose firm, Kobre & Kim, would go on to represent Sarao and several other parties accused of spoofing. ‘There are no rules on how long you have to leave an order up so the action itself is totally legal. It’s only the state of mind that matters. The whole thing has turned into this bizarre thought crime.’

  After Dodd-Frank was signed into law in July 2010, the Disruptive Trading Practices amendment underwent a process of refinement until, three years later, the CFTC issued a document containing ‘interpretive guidance’ on what was and wasn’t allowed. It stated that spoofing included intentionally submitting and canceling orders to give ‘an appearance of false market depth’ or ‘create artificial price movements’. To differentiate between acceptable and unacceptable activity, the agency said it would take into consideration ‘the market context, the p
erson’s pattern of trading activity … and other relevant facts and circumstances’, giving credence to one commentator’s suggestion that spoofing was like pornography – you knew it when you saw it.

  There is a common misconception, particularly among day traders, that HFT is responsible for most of the spoofing that goes on in financial markets. While it’s true that several HFT employees and firms have been sanctioned for engaging in the practice over the years, for the most part it’s actually looked down on by the behemoths at the forefront of the industry. For them, spoofing is a hindrance, and after the introduction of the disruptive trading rules, firms such as Citadel and HTG Capital Partners would go on to work closely with regulators to identify and prosecute parties whose unpredictable trading disrupted the smooth functioning of their moneymaking machines. In a number of cases they appeared as witnesses for the government. ‘Imagine a system where the biggest, most powerful players get to tell the regulators and the exchanges who to go after based on who is taking money off them,’ says one prop trader. ‘Welcome to the futures market.’

  CHAPTER 15

  PIMP MY ALGO

  In October 2011, Nav was hit with a more immediate threat to his livelihood than a change in the trading rules. MF Global, the firm that gave him access to the markets and provided him with the leverage to trade at such high stakes, declared itself bankrupt. Problems had been swirling around the world’s biggest and oldest futures broker for a while, but few realised the matter was terminal, and when MF Global filed for Chapter 11 protection on Halloween, tens of thousands of traders, Nav among them, were left without access to their funds and frantically searching for a new home.

  Brokers make their money in two principal ways: from the fees they charge customers per trade and on the returns they make investing the excess funds sitting in client accounts. When the financial crisis struck, both revenue streams dried up as trading volumes plummeted and central banks cut interest rates to the floor. In March 2010, after three years of losses, MF Global’s board appointed a new CEO to turn things around. Their pick, Jon Corzine, was a bona fide Wall Street legend who had run Goldman Sachs in the nineties before cashing in his considerable chips and entering politics, first as a senator then as governor of New Jersey. ‘It was as if a manager of the New York Yankees was making a comeback in the minor leagues,’ wrote the New York Times.

  Corzine, who was a spry sixty-four, had started his career as a bond trader, and he believed the solution to the firm’s problems, at least in the short term, lay on the trading floor. It was the middle of the Eurozone crisis, and bonds issued by the likes of Italy, Spain and Portugal were trading at considerably below their face value, reflecting the heightened probability of default. Convinced the authorities would never let that happen, Corzine quietly bought up $7 billion of the distressed securities, using them as collateral for a complicated so-called ‘repo’ trade that served the dual purpose of keeping the instruments off the firm’s balance sheet and allowing it to declare its expected profits straightaway. It was the kind of bold, outside-the-box play that had catapulted Corzine from the family farm in Illinois to the Bilderberg Group. Unfortunately, this time the former high school quarterback had badly misjudged his Hail Mary pass. As the crisis deepened and the value of sovereign debt fell, the counterparties on the repo trades began issuing what are known as margin calls, forcing MF Global to hand over additional cash to cover potential losses. In October 2011, rating agencies downgraded the firm’s credit rating to ‘junk’, sparking the equivalent of a bank run, with customers calling up and demanding to withdraw their funds and lenders closing credit lines. In the frenzied final hours, a middle-ranking executive dipped into what should have been sacrosanct, segregated client funds to plug the gaps. It wasn’t enough to save the firm, and in the days after MF Global folded, it emerged that $1.6 billion belonging to twenty-six thousand retail traders, farmers and small businesses was unaccounted for.

  While Corzine faced uncomfortable questions about his level of involvement in the scandal, Nav considered a future without his long-standing broker. Financially, thanks to the offshore guarantee arrangements, the blow was manageable. Nav had only a modest sum sitting in an MF Global margin account, most of which he’d eventually get back. Of greater concern was finding a new broker. GNI Touch, the MF Global offshoot, had watched Nav develop; they’d grown comfortable with his trading style and the size of his positions. In exchange for a steady torrent of commissions, the firm’s brokers had given Nav the best terms and stayed out of his way even when his methods drew unwanted attention. Finding such a generous and pliant replacement would prove difficult.

  The first problem was Nav himself. By any measure, his record was stellar, but few clearers were willing to back someone with millions of pounds without meeting them first, and the eccentric trader didn’t always make the best impression. After calling one City brokerage and requesting an after-hours appointment so no one would see him, Nav turned up wearing sweatpants and carrying his trading statements in an old Tesco plastic bag. ‘He was in the office for about an hour. He wouldn’t look you in the eye. You had to pull information out of him. I was about as unimpressed as it’s possible to be,’ recalls one of the executives who interviewed him. Still, day traders are an odd bunch and, recognising there was money to be made, the firm offered Nav a contract on less attractive terms and lower leverage than he was looking for. ‘We talked about it and figured he would probably carry on making good money, but if he really hammered it at the levels he wanted and the market jumped back and he got caught, he could have wiped us out,’ says the executive. Nav declined and walked away. ‘He wanted to be worshipped,’ the firm’s owner says. ‘Not only did he want outsized risk at frankly uneconomical terms, but he wanted us to ruffle him, which we weren’t prepared to do.’

  The monumental size of Nav’s positions was another issue. As he later recounted to a friend: ‘I tried to get a broker’s deal with RCG,’ the Rosenthal Collins Group, one of the biggest clearers in the United States. ‘They said, “How much do you make?” and I said, “On a good day, nine hundred grand.” They said, “That’s crazy.”’ A week later, after Nav sent them his statements, RCG turned him down. When he asked why, they told him they didn’t think it was possible for a trader in his bedroom to make that much money. ‘They said they believe that it’s a Ponzi scheme,’ Nav recounted. ‘They thought I was Bernie Madoff!’ Nav hoped he’d found an alternative when Knight Capital agreed to take him on, but then the firm lost $460 million and he was left brokerless once again.

  Unable to trade and with few other interests, Nav returned to the proverbial garage to work on his trading machine. The layering algorithm he’d created in 2009 with help from Hadj at Trading Technologies had worked exactly as planned, firing blocks of sell orders into the market to drive prices lower that stayed far enough from the prevailing price to almost never be hit. But it was a blunt instrument, and Nav’s competitors – principally the big HFT firms and a handful of gifted manual scalpers like him – were getting stronger and more sophisticated by the day. To bait the latest algos, he needed to be able to place his spoofs nearer the action, either at or close to the best offer, and that was risky. Nav knew he had to evolve. So on a Monday evening in October 2011, he left a voicemail with a software developer in Chicago who specialised in pimping TT systems.

  AROUND THE time Jitesh Thakkar listened to Nav’s message, he was engrossed in a book called Secrets of the Millionaire Mind by T. Harv Eker, a mullet-sporting fitness store owner turned motivational speaker. Thakkar had his own business and was a sucker for a self-help book. ‘Rich people believe: “I create my life.” Poor people believe: “Life happens to me,”’ he wrote in a fifteen-point summary of Eker’s bestseller that he posted on social media. ‘Rich people think big. Poor people think small.’ Thakkar had already devoured The Secret, Rhonda Byrne’s touchstone on the law of attraction. ‘Ask. Believe. Receive,’ he repeated to himself on the commute from suburban Napervi
lle to his office downtown.

  Thakkar’s family had moved from India to Chicago when Jitesh was thirteen years old, and his path in life was set a few months later when he came across an Apple model IIe in the computer lab at school. Beige and boxy, with a green and black screen, it allowed users to write programs and play rudimentary games. Thakkar was smitten and taught himself to code. He studied computer engineering at the University of Illinois, where, in his spare time, he designed chess and puzzle games in his bedroom. One of Thakkar’s first jobs was at the investment bank UBS, where he created applications to price options. In 2001, he took a role at a market maker called Stafford Trading, building systems that cut down any lag between a customer placing an order and its hitting the exchange. It was a much sought-after skill, and two years later he was hired by Trading Technologies to oversee a department. Away from work, Thakkar and his wife were adherents of The Art of Living, a spiritual programme based around the teachings of a guru named Ravi Shankar. The couple gave classes on meditation and deep breathing and went to ‘Yoga Raves’, where sober crowds danced and whooped to Indian-tinged electro music.

  In 2007, with the HFT industry exploding, Thakkar left TT to start his own firm, Edge Financial Technologies. As well as helping customers minimise latency, Edge introduced old-school point-and-click traders to automation by working with them on designing and developing their own algorithms. He rented an office near the CME and assembled a small team of developers. There weren’t a lot of other consultants targeting TT users – most sizable HFT firms built their own software systems from scratch – and business flowed in via referrals from his former employer. ‘Jitesh was the guy,’ says a former TT employee. ‘If a customer wanted anything a little bit complex, anything beyond what we could do ourselves, we sent them his way.’