Flash Crash

by Liam Vaughan

  The problem with spoofing as a strategy was that it was somewhat akin to picking up pennies in front of a steamroller. To alter the order book sufficiently to have an impact, you had to place big orders. But if a massive hedge fund or bank happened to come along at the wrong moment with a mandate to hoover up a billion dollars of e-minis and lifted all your resting offers, you’d be left watching in horror as the market jumped ten levels, costing you millions. Placing orders several ticks from the prevailing price, as the layering algorithm was designed to do, was pretty safe. But Nav knew if he really wanted to incite the algos, it would be useful to be able to also place some spoofs at or near the best offer, and that was dangerous. The solution he came up with was called ‘back of the book’, and it was designed to take advantage of the CME’s First In, First Out (FIFO) queuing system.

  Imagine that the e-mini market is a supermarket, and each price level on the ladder represents a different register. Every time a fresh order is placed at a certain price, regardless of its size, it joins the back of the queue for that price and moves steadily further forward as the price fluctuates and the orders in front of it are matched. If, however, a participant adds to their order, like a customer stepping away for a moment to add to their cart, they’re judged by the CME to have left the line and are sent to the back. Nav’s brainwave was to have his algo add a single lot every time a fresh order arrived behind him, thereby constantly sending him to the back of the line and out of harm’s way. To keep his order size at the amount he intended, the algo would subtract a solitary contract the next time an order arrived, alternating between plus one and minus one in perpetuity. As a further safeguard, Nav proposed incorporating a feature he called ‘one clip’, which dictated that if any portion of his orders was inadvertently hit, the remainder would immediately be cancelled, thereby limiting any potential damage.

  Nav’s email was forwarded to an engineer in New York named Antonios Hadjigeorgalis, whose job was to help customers get the most out of the Autotrader system. Hadj, as he was known, was a frustrated trader of Greek descent who had taken the job at TT in 2007 when his trading hit a dry patch. In his spare time he was a devotee of self-improvement who practised yoga and martial arts, followed an ultralow-carb diet and kept a blog where he reviewed the hundreds of books he speed-read each year. The team Hadj joined was badly overstretched. When he arrived there were four engineers globally, but soon his counterpart in London quit and Hadj was left to pick up the slack. That year he clocked up more than a hundred thousand air miles.

  On a summer day in 2009, Hadj touched down in London for a busy schedule of meetings with existing and prospective clients organised by the sales rep. At CFT’s offices, Nav talked through the features he wanted. ‘Cancel if close’ was straightforward to set up on the Autotrader system, and Hadj had it up and running within a few minutes. ‘Back of the book’ was more unusual. HFT firms spent millions of dollars to make their systems faster so they could get to the front of the queue, but Nav wanted something that did the opposite. Hadj promised to look into it and they parted ways. A couple of months later, in November, Nav emailed again. ‘The system you set up … whereby I turn the Autotrader on or off and when it was turned on it would put offers a specific value and quantity away from the best offer’ was proving ‘really useful,’ he wrote. ‘I remember you typing in a code to enable this and was wondering whether you could tell me what it was so I could play around with creating new versions of the same thing.’ Hadj agreed, but over the next few weeks Nav bombarded him with additional requests until the engineer eventually grew frustrated. All TT customers paid the same monthly fee and any assistance they received was discretionary, but Nav acted as if Hadj was working for him. When Nav pressed him about ‘back of the book’ again, Hadj told him it was beyond Autotrader’s functionality and he would need to find an external programmer. Sarao’s insistence struck Hadj as odd, but before long TT hired a new engineer in London and it was no longer his problem. He wouldn’t give Navinder Sarao another thought until six years later when he saw him on the evening news.

  CHAPTER 10

  THE CRASH

  On Thursday 6 May 2010, Nav awoke in his upstairs bedroom, got up and switched on the computer that sat at the end of his single bed. He still kept a desk at CFT, but he preferred to trade from his parents’ home in Hounslow where there were no distractions or prying eyes. His was a solitary and nocturnal existence. He had few close friends and, beyond the occasional game of snooker or kick-about in the playing fields near his house, he barely left the house. His parents, Nachhattar and Daljit, pressed him to find a wife, but trading was his passion. Every Sunday they headed to the temple with their devout, turban-wearing eldest son Rajvinder and his young family while Nav stayed home and slept.

  It’s difficult to imagine what it must feel like to invent a machine that prints money and then not tell anyone about it, but that’s essentially what Nav had done. The program he’d made using TT software hadn’t just worked. It had proven wildly, scarily effective. Following some teething problems, Nav had fine-tuned the system to the point where he could more or less nudge one of the world’s biggest markets around at will. The previous day, over a few hours, he’d made $435,185, more than the value of his parents’ house. The day before that, his profit was $876,823 – seven times what his hero, Lionel Messi, earned per day at FC Barcelona. Early in his career, Nav made a decision not to talk to his friends and family about his finances because he was worried they would treat him differently. Now, at thirty-one years old, he was outearning the highest-paid footballer in the world, and almost nobody aside from his brokers and a couple of financial advisers knew about it.

  Nav’s activity hadn’t gone entirely unnoticed. A few weeks earlier, on 23 March a member of the CME’s in-house surveillance unit in Chicago emailed GNI Touch to notify it that, in the space of five minutes, its client had had 1,613 trades rejected with the message ‘This order is not in the book’. Competing market participants had evidently seen Nav’s orders resting on the ladder but, when they tried to hit them, were informed that they had disappeared. GNI looked into the issue and determined that Nav was using his software to delete ‘a huge amount of orders a second’. They forwarded him the CME’s message and advised him to cut it out.

  The following week, Nav emailed the CME, cc’ing his brokers, to ‘apologise for any inconvenience caused by this’. He said he ‘was just showing a friend of mine what occurs on the bid side of the market almost 24 hours a day, by the high frequency geeks’, and asked whether the CME’s interest in him meant that ‘the mass manipulation of the high frequency nerds’ was also going to end. Then he went straight back to using the program.

  Outside, in the real world, it was a general election day, and residents made their way to polling stations to cast their vote. Britain was in tumult after the financial crisis. Unemployment was up 50 per cent, banks like RBS and Lloyds had been nationalised, and the economy remained on life support. The left-leaning Labour Party, which had swept to power on a wave of optimism in 1997, was facing defeat by the centre-right Conservatives. Nav didn’t care too much either way. As far as he was concerned, all politicians were equally bad. But political uncertainty fed into the markets, which was good for trading.

  Nav’s trading strategy depended on volatility, and he monitored conditions closely, like a surfer waiting for the perfect swell. When seas were flat, he stayed away. The last couple of weeks had been phenomenal. Between mid-2009 and April 2010, global stock markets had bounced back from their post-crisis lows thanks to interest rate cuts and a glut of central bank money. But the crisis had exposed something rotten in the European Union, and the smell was getting stronger. After bailing out their domestic banks, a number of countries, particularly the so-called PIIGS (Portugal, Italy, Ireland, Greece and Spain), were saddled with debt and battling recession, unemployment and social upheaval. The biggest basket case was Greece, where the situation had deteriorated so much it could no longer afford
to service its debts. The country was effectively bankrupt, and on 2 May the European Commission, the European Central Bank and the International Monetary Fund announced they would provide a $145 billion lifeline. In return, Greece’s leaders agreed to implement a programme of severe public sector cuts. It was the final straw for an already beleaguered population, and on 4 May, thousands of protesters stormed the Acropolis in Athens, a stark symbol of how far the country had fallen.

  The buzzword in the press was ‘contagion’. If countries started defaulting, the banks that held their debt would require state aid and would be unable to invest in future government bond issuances, leading to a death spiral. And propping up a large economy like Italy or Spain was a different proposition to Greece. With the future of the EU hanging in the balance, institutional investors scurried to safety, pulling assets out of Eurozone sovereign bonds and stocks and piling into gold and Treasury bills. By the morning of 6 May, the ‘fear index’, a measure of expected volatility in the S&P 500, was up 16 per cent on the start of the week.

  Nav bided his time until 3:20 p.m. – 9:20 a.m. in Chicago – when, with a click of the mouse, he switched the Autotrader program on. Despite his recent windfall, his home setup was no more complex than the pared-back arrangement at Futex: three monitors containing ladders, charts, a news feed and a TT interface; a standard keyboard; and a mouse. The only sounds were the planes passing overhead and the thrum of a PC fan straining to combat overheating. Nav activated the ‘cancel if close’ feature and placed four sell orders totaling 2,100 contracts one tick apart, starting three levels above the best offer of 1,163.25. They had a combined value of $120 million. Over the next six minutes, as the e-mini price fluctuated, these orders were automatically cancelled and replaced 604 times to ensure they remained in lockstep and therefore unconsummated. With markets already falling, and so much other trading going on, it’s almost impossible to know exactly what impact Nav’s spoofing had, but by the time he clicked the algo off for the first time that day, the market had tumbled 39 points.

  As the day progressed, a pervasive feeling of anxiety took hold of the markets. Apocalyptic news reports from Greece showed black-clad demonstrators hurling Molotov cocktails at armoured police, who struggled to keep them at bay with water cannons. In Lisbon, an obdurate Jean-Claude Trichet, head of the European Central Bank, ruled out adopting more drastic measures to curb the crisis, pushing Spanish bond yields to a twelve-year high and forcing the euro lower. By the European close, the EURO STOXX 50 index of blue chips was down 3 per cent. The S&P 500 was close behind. If, as Nav had told his acolytes at Futex, the markets were no more than a massive psychological barometer, then the dial was teetering somewhere between fear and panic.

  It was on days like this that Nav’s approach of focusing his trading on the short side of the market made most sense. When markets rise, they tend to do so in an orderly fashion, climbing steadily over weeks and months. But when they fall, the correction can happen very quickly, and when that occurred, Nav wanted to be positioned to reap the rewards. It explained why, as he would later tell the regulators, he made the bulk of his wealth in no more than twenty days spread across his trading career.

  A couple of hours after sitting down to trade, at 5.17 p.m. London time, Nav activated the Autotrader for the last time that day. Usually he liked to use it in five- or ten-minute cycles, perhaps to evade detection or manage his exposure, but on this occasion he left the orders sitting there for more than two hours. He also cranked up the size. Nav started the cycle by placing five sell orders of six hundred lots each, three, four, five, six and seven ticks above the best offer. But as the sky grew darker outside, he added a sixth, bringing the total value of his spoof offers to $200 million. That barrage sat in an order book that was already severely imbalanced, helping push the number of sell orders to twice the volume of resting bids. To ramp up the pressure even further, Nav used his mouse and keyboard to intermittently flash scores of additional spoofs of 289 and 188 lots. As the market tumbled, he nimbly carried out his genuine trades, selling chunks of e-minis at a time, then buying them back for a few ticks less.

  Eventually, at 7.40 p.m. in London, 1.40 p.m. in Chicago, Nav closed the system down and stopped trading for the day. Why he chose to stop at that exact moment is not clear. Maybe his mother called him down for dinner. It was, regulators would later calculate, the second-most-frenetic trading session he would ever have, involving somewhere in the region of 18.5 million orders. In that final two-hour spell alone, he’d bought and sold 62,077 e-mini contracts with a combined value of $3.4 billion. If, at any moment, the market had rallied, his entire account could have been wiped out. Instead, the e-mini fell 361 points, and he made a profit of $879,018. What happened next, he was only a spectator for.

  One minute after Nav shut off the Autotrader program, at 1.41 p.m. CT, the e-mini started to plummet with a velocity and intensity it never had before. He watched as the S&P 500 shed 5 per cent of its value in four minutes, more than it had throughout the entire day to that point, creating a price chart that resembled a cliff face. Almost simultaneously, the closely correlated SPDR exchange-traded fund (known as ‘the Spider’) followed suit on the New York Stock Exchange. Next, individual shares began to tumble, igniting a sea of flashing red on traders’ screens around the world. The fear index barrelled 20 then 25 then 30 per cent higher as the Dow hurtled below 10,000 points for the first time in six months. Terrified market participants pulled the plug, draining liquidity and causing the e-mini to cascade lower in gallops of five and ten ticks at a time. At one stage, even crude oil started sinking. From Frankfurt to Shanghai, interconnected financial markets went haywire, an apocalyptic nightmare scenario playing out in fast-forward. Then, at exactly 1.45 and 28 seconds, the e-mini ladder froze. The CME’s ‘stop-logic’ function had kicked in after the rate of the fall had breached a set level and, for five long seconds, no trading took place. Around the world, human traders and algorithms paused in unison. The Dow had fallen more in a five-minute period than at any other time in its 114-year history.

  When trading resumed, the e-mini started to climb as rapidly and miraculously as it had tanked. The S&P 500 jumped from a low of 1,056 at 1.45 to 1,096 at 1.50 and 1,120 three minutes later, transforming the e-mini chart into a steeply sided V shape. Traders hadn’t suddenly woken up from a collective stupor and realised that prices were out of whack. Transactions were occurring automatically at hyperspeed, algorithms interacting with algorithms in frenzied and unpredictable ways. Even as some participants rushed to the exit, the volume of transactions spiked to near-all-time highs, as futures passed back and forth like pinballs.

  Away from the CME, the day’s most bizarre events were still to come. Between 1.45 and 2.00 p.m. CT, shares in some of America’s most familiar corporations changed hands at prices utterly divorced from anything resembling fair value. Proctor & Gamble, Hewlett-Packard, General Electric and 3M plummeted 10 per cent or more, while the iShares Russell 1000 Value Index, a popular exchange-traded fund, fell from $50 to 0.0001 cents. Accenture sold for a solitary cent. At the other end of the spectrum, Apple and auctioneer Sotheby’s both transacted at $100,000 a share, momentarily pushing their valuations into the trillions of dollars.

  The stock market’s twilight zone would also prove short-lived. As the e-mini continued its bounce back, participants tentatively returned to equities markets and individual shares began trading again at levels close to where they’d been before 1.30 p.m. Within half an hour, markets had retraced the bulk of their losses, and by the time NYSE closed, the Dow was back at 10,520.32 points, a sizable but unremarkable 3.2 per cent decline on the day. Any trader who happened to leave his desk at 1:30 p.m. on 6 May 2010, and grab a cup of coffee would have missed it, but for twenty minutes or so, the financial world had stared into the abyss. In the end, for reasons nobody quite understood, disaster had been averted, but the repercussions would play out for years to come. As for Nav, he took a couple of days off. Later in the mon
th, when the CME sent him another reminder of the requirement to enter transactions in good faith, he emailed his broker. ‘Just called the CME,’ he wrote, ‘and told em to kiss my ass.’

  ACT TWO

  CHAPTER 11

  THE AFTERMATH

  The US Treasury secretary, Tim Geithner, scheduled an emergency call of the President’s Working Group on Financial Markets for 6.30 p.m. Eastern time on 6 May 2010, after the markets had closed. Joining him on the line were Gary Gensler and Mary Schapiro, heads of the Commodity Futures Trading Commission and the Securities and Exchange Commission, respectively; Ben Bernanke from the Federal Reserve; Bill Dudley from the New York Fed; and all the other leaders of the country’s major financial authorities. Between them, they were responsible for markets worth hundreds of trillions of dollars. The group had been created by Ronald Reagan in the aftermath of the October 1987 crash to facilitate better coordination between government agencies. Colloquially, it became known as the ‘Plunge Protection Team’ thanks to an unsubstantiated but pervasive rumour that it routinely intervened in markets to serve government ends. The purpose of the call this particular evening was to find an answer to a question being asked by millions of people around the world at that moment: What the fuck just happened?