Flash Crash 2010: The Day the Machines Went Rogue | Market Mayhem EP16

At 2:32 PM on May 6th, 2010, a single algorithm began executing a $4.1 billion sell order in the US futures market. No fraud. No conspiracy. No rogue actor with malicious intent. A standard institutional hedging strategy.

Fifteen minutes later, the Dow Jones Industrial Average had fallen 998.5 points — its largest intraday point decline in history. Accenture, a company worth tens of billions of dollars, traded at one cent per share. Procter and Gamble fell thirty-seven percent. Nearly one trillion dollars in market value had evaporated.

Twenty minutes after the bottom, almost all of it had returned.

A trillion dollars. Gone and back in thirty-six minutes. The purest demonstration of what modern market structure can produce when the algorithms that provide liquidity all make the same decision at the same moment.

---

The Crisis at a Glance

Data Point	Detail
Event	Flash Crash — the largest and fastest intraday point decline and recovery in US stock market history
Date and Time	May 6, 2010 — 2:32 PM to 3:07 PM (approximately)
Dow Jones Decline	998.5 points intraday — the largest intraday point drop in history at that time
Time from Peak Decline to Recovery	~20 minutes — most losses recovered within the same trading session
Market Value Lost at Trough	~$1 trillion — briefly, before recovery
Accenture Price	Traded at $0.01 per share briefly — from a normal price of approximately $40
Procter and Gamble Decline	37% decline in minutes, before recovering
Triggering Order	Waddell and Reed mutual fund — $4.1 billion in S&P 500 futures, executed via volume-based algorithm
Amplifying Mechanism	High-frequency trading firms withdrawing liquidity simultaneously as conditions deteriorated
Navinder Sarao Contribution	Spoofing in S&P futures from a London suburb — arrested 2015, sentenced to time served; contributed to but did not solely cause the crash
Regulatory Response	Individual stock circuit breakers; Limit Up-Limit Down mechanism; enhanced market surveillance for spoofing
M·M·M Lesson	Method — the market you think you’re trading is not the market you’re actually trading. Money — stop-loss orders and flash events. Mind — complex systems produce emergent behaviours that cannot be fully predicted.

---

The Market Nobody Fully Understood

The US stock market of 2010 was radically different from the image most people held of it. The New York Stock Exchange trading floor — specialists in coloured jackets, open-outcry trading — was largely television set-dressing. The real market was electronic, fragmented across approximately fifty trading venues, and operating at speeds measured in microseconds.

At the heart of this system were high-frequency trading firms — technology companies using co-located servers and proprietary data feeds to trade at microsecond speeds. In normal markets, they provided genuine value: narrower bid-ask spreads, faster execution, cheaper transaction costs for ordinary investors. They had transformed the market from an expensive, slow institution of the 1990s into a cheap, fast one by 2010.

The structural feature that mattered on May 6th: this liquidity was conditional. High-frequency traders had no obligation — unlike traditional designated market makers — to provide two-sided quotes regardless of conditions. They provided liquidity when it was profitable. When conditions deteriorated, they could withdraw instantaneously. And in a market where their liquidity provision was the foundation of normal price formation, their simultaneous withdrawal created a vacuum.

The Sequence: 15 Minutes That Changed Market History

2:32 PM: Waddell and Reed’s volume-based algorithm begins selling $4.1 billion in S&P 500 futures. The market is already down ~3% on Greece news. Volume is elevated, which means the algorithm sells faster than in normal conditions.

~2:41 PM: The selling has absorbed roughly a third of the order. Prices are falling. High-frequency trading firms, detecting sustained selling and deteriorating price action, begin pulling their bids. As bids disappear, the remaining sell orders hit a market with almost no buyers. Prices move further, prompting further bid withdrawal. The liquidity spiral is running.

~2:45–2:47 PM: Procter and Gamble falls 37%. Accenture trades at one cent. The Dow is down nearly 1,000 points in approximately fifteen minutes. The market’s price formation mechanism has essentially broken.

~2:48 PM: The algorithm completes most of its selling. Without the continued selling pressure, high-frequency firms re-enter. Bids reappear. The market finds its footing and reverses almost as rapidly as it fell. By 3:07 PM, most of the loss has been recovered.

The SEC/CFTC investigation that followed took months and hundreds of pages to confirm what the timeline suggested: the triggering event was the algorithm; the amplifying mechanism was the simultaneous liquidity withdrawal; the recovery was the restoration of that liquidity once the pressure ceased. No individual had decided to crash the market. The crash had emerged from the interaction of individually rational systems.

Navinder Sarao: The Spoofer in the Suburb

Five years later, a second explanation emerged. Navinder Singh Sarao — trading from his parents’ house in Hounslow, near Heathrow — was alleged to have been spoofing S&P 500 futures for years. Spoofing involves placing large orders with no intention of executing them, to create false impressions of supply or demand that move prices to benefit your actual positions.

On May 6th, 2010, Sarao’s spoofing — according to the DOJ — had added to the selling pressure already created by the Waddell and Reed algorithm, contributing to the conditions that triggered the crash. He pleaded guilty and cooperated with authorities, receiving a sentence of time served. He had reportedly made approximately $40 million over years of spoofing — much of which had been lost to fraudsters who defrauded him.

Whether Sarao was the primary cause or a contributing factor remains debated. The honest assessment: both the algorithmic trigger and his spoofing were probably necessary; neither was sufficient alone; and the market structure that allowed either to trigger a trillion-dollar price movement in fifteen minutes is the deeper problem.

---

What This Means for You as a Trader

---

Frequently Asked Questions

What is spoofing and why is it illegal?

Spoofing involves placing orders that a trader has no intention of executing: large buy or sell orders placed in the order book to create a false impression of demand or supply, which moves the price in a direction that benefits the trader’s actual positions, at which point the spoof orders are cancelled. It is illegal under the Dodd-Frank Act (in the US) and equivalent legislation elsewhere because it constitutes market manipulation. It deceives other market participants about genuine supply and demand conditions, leading to price formation that does not reflect real economic interest. Sarao’s conviction was one of the highest-profile spoofing prosecutions, and enhanced market surveillance technology since 2015 has made spoofing significantly harder and the consequences significantly more severe.

What is high-frequency trading and is it good or bad for markets?

High-frequency trading (HFT) is the use of sophisticated algorithms, co-located servers, and proprietary data feeds to trade financial instruments at speeds measured in microseconds. The debate about whether HFT is beneficial or harmful is genuine and unresolved. On the beneficial side: HFT firms have dramatically narrowed bid-ask spreads and reduced transaction costs for all investors; they provide continuous liquidity in normal conditions; they contribute to price efficiency by rapidly arbitraging away price discrepancies across venues. On the harmful side: the liquidity they provide is conditional and can disappear exactly when it is most needed; they create a two-tier market where the fastest participants have structural advantages over ordinary investors; and, as the Flash Crash demonstrated, their simultaneous behaviour in stressed conditions can amplify price movements dramatically. The honest answer is that HFT is genuinely beneficial in normal conditions and a genuine risk factor in extreme ones.

Could the Flash Crash happen again today?

The specific conditions of May 6th, 2010 — no individual stock circuit breakers, a different market fragmentation pattern, Sarao’s specific spoofing activity — are less likely to recur in exactly that form. However, the underlying structural feature, conditional liquidity from high-frequency traders that can withdraw simultaneously in stressed conditions, has not been eliminated. Mini flash crashes in individual stocks continue to occur periodically. A market-wide event of Flash Crash magnitude is constrained by circuit breakers, but the fundamental market microstructure vulnerability that enabled it is a permanent feature of algorithmic markets. The next major flash event will probably have a different trigger, a different affected market, and a different specific mechanism, but the same underlying structural DNA.

How did circuit breakers fail to prevent the Flash Crash?

In 2010, circuit breakers existed at the market level (a market-wide halt triggered by large percentage declines in major indices) but not at the individual stock level. The Flash Crash’s most extreme manifestations were in individual stocks: Accenture at one cent, Procter and Gamble down 37%. These individual stock moves were not covered by the existing circuit breakers. The new individual stock circuit breakers introduced after the Flash Crash, and the later Limit Up-Limit Down mechanism, are specifically designed to prevent individual stocks from trading at prices more than a specified percentage from a reference price. They have been effective in preventing the most extreme instances of Flash Crash-type price moves in individual securities.

What should a retail trader do differently after understanding the Flash Crash?

Several practical changes: use limit orders rather than market orders where possible, particularly in volatile conditions or when trading less liquid securities, to ensure you do not execute at dramatically worse prices than intended. Understand the mechanics of your stop-loss orders, whether they are market stops (execute at any price once triggered) or limit stops (will not execute below a specified price). Be aware that in extreme market events, the price displayed on your trading platform may lag actual market prices or may reflect trades at non-representative prices. And perhaps most fundamentally: size your positions such that even a Flash Crash-type event in your position does not cause terminal damage to your portfolio. If a 37% move in a single holding is catastrophic for your account, you are holding too much of it.

---

Market Mayhem is a historical education series produced by The Complete Trader’s Edge. All figures are sourced from historical records. Content is for educational purposes only and does not constitute financial or investment advice. Trading involves significant risk of loss.