● A free public tool  ·  Built by Poptani Labs  ·  No sign-up  ·  Your data never leaves your browser
The Butterfly Effect · 5-minute read

The 2010 Flash Crash: how $1 trillion vanished in 36 minutes

Traders working on the floor of the New York Stock Exchange
Image: 'NY stock exchange traders floor LC-U9-10548-6.jpg' by Thomas J. O'Halloran, Public domain, via Wikimedia Commons

On Thursday, May 6, 2010, at 2:32 p.m. Eastern Time, the US stock market began to fall — and then it fell off a cliff. In about fifteen minutes the Dow Jones Industrial Average plunged a further 600 points, taking its loss for the day to 998.5 points, about 9 percent. Press coverage at the time described more than $1 trillion in market value vanishing in the process. By 3:07 p.m., most of it had come back. The whole episode lasted about 36 minutes. No company had gone bust. No bomb had gone off. The cause was plumbing: a single sell order, an algorithm with no brakes, and a market that had quietly run out of buyers.

The order that started it

The trigger was a mutual fund company in Kansas called Waddell & Reed. It held a large position in American shares and wanted to hedge it, so it began selling 75,000 E-mini S&P 500 futures contracts — roughly $4.1 billion worth. The trade was handed to a computer program with one instruction: sell at a rate of 9 percent of whatever the trading volume had been over the previous minute. The program was given no limit on price and no limit on time. It was told only to match volume. It completed the entire sale in approximately 20 minutes — a pace that would normally require five hours or more for a block of that size, according to the joint CFTC-SEC report issued in September 2010. And it created a vicious loop: as the selling pushed prices down, trading volume rose, and the algorithm — told to keep pace with volume — sold faster still. The market was already nervous that day, down on worries about the debt crisis in Greece, protests in Athens, and elections in the UK. The algorithm did not know that. It did not know anything at all.

The market with no buyers

As prices fell, the high-frequency trading firms that normally supply the market with buyers and sellers did what their systems told them to do: they paused, scaled back, or withdrew altogether. The market makers who stayed behind posted what are called stub quotes — placeholder bids of a cent, or offers of $100,000, prices so absurd that nobody was ever meant to hit them. With no real buyers left, the placeholders became the market. Accenture, a company worth tens of billions of dollars, traded at $39 — and then, trade by trade, at $32.62, $5.34, $4.04, $1.84, and finally one cent, before closing the same afternoon at $41.09. Sotheby's, which had been trading around $34, printed at $99,999.99. Procter & Gamble, one of the steadiest shares in America, fell from around $60 to $39.37 in about three and a half minutes. At 2:45 p.m., a built-in safeguard at the Chicago Mercantile Exchange — the stop-logic function — triggered a five-second trading pause in the E-mini contract after detecting the velocity of the decline, and the market began to stabilise. The exchanges later cancelled trades in 286 different securities as “clearly erroneous.”

The trader in the bedroom

Five years later, the story took a stranger turn. In April 2015, the US Department of Justice arrested Navinder Singh Sarao, a 36-year-old British trader, on 22 counts including fraud and market manipulation. Sarao worked from his parents' house in Hounslow, west London, using commercially available trading software whose code he had modified so he could rapidly place and cancel orders automatically. His technique was spoofing: placing thousands of E-mini orders he never intended to execute, to create a false impression of supply and demand. On the morning of the crash, his orders — amounting to about $200 million in bets that the market would fall — were replaced or modified 19,000 times before they were cancelled. In 2016 he pleaded guilty to spoofing and wire fraud, and in 2020 a federal judge sentenced him to home detention rather than prison. The CFTC concluded he was “at least significantly responsible” for the order imbalances that day — though how much of the crash was his doing remains fiercely contested. The joint CFTC-SEC report had already blamed the Waddell & Reed algorithm and the withdrawal of liquidity as the main causes.

What changed

The crash rewrote the rules. Spoofing, layering, and front running were banned under the Dodd-Frank Act, and the exchanges formalised common procedures for cancelling “clearly erroneous” trades. The deeper fix was about speed itself: markets had become so fast that a single badly designed algorithm could move the entire system in minutes, and the safeguards were built for a slower world. The Flash Crash is now the textbook case of how that happens — not one mistake, but a chain of them: a blunt algorithm, a nervous market, and a liquidity system that vanished exactly when it was needed most.

The Lesson

The Flash Crash lasted 36 minutes and changed nothing about any company in the market — and yet it changed the market itself. The lesson is the one the Butterfly Effect always teaches: the failure that kills a system is rarely one big mistake. It is two or three small ones, arriving at the same moment, in a system that assumed they never would.

Related on Poptani Labs: The missing hyphen that blew up a rocket to Venus, The butterfly ballot: how a confusing ballot design decided a presidency, The officer who ignored five nuclear missiles — and saved the world. More in this category: The Butterfly Effect.

Sources: Wikipedia, 2010 flash crash, Market Histories, The Flash Crash: When Algorithms Broke the Market in 36 Minutes (2010), Glenmede, The Flash Crash: Are the Markets Rigged?

← All posts