2010 Flash Crash: When Liquidity Disappeared in Minutes

Key Takeaways

• The Dow fell roughly 1,000 points intraday on May 6, 2010, with individual stocks briefly printing at one cent or $100,000 before recovering in under 30 minutes.
• A single algorithm selling 75,000 E-Mini S&P futures worth $4.1 billion, without price or time limits, triggered a "hot potato" chain reaction among high-frequency traders.
• Investors using naked stop-loss orders can receive fills at catastrophic prices during liquidity voids; stop-limit orders constrain the worst outcomes.
• Quoted depth in electronic markets is not real resilience, market makers can pull quotes in seconds, collapsing apparent liquidity to nearly zero.

What It Is

On the afternoon of May 6, 2010, US equity markets were already down on the day when the drop suddenly accelerated around 2:40 PM Eastern Time. The Dow fell roughly 1,000 points intraday, about 9%, reaching its low near 2:47 PM. Prices then rebounded sharply, closing down 3.2% on the day.

During the worst minutes, individual stocks printed at nonsensical prices. Accenture traded at one cent per share. Some other shares briefly printed at $100,000. The exchanges later cancelled trades that were executed more than 60% away from the reference price immediately before the crash.

The SEC-CFTC joint report, released September 30, 2010, identified a large automated sell program in E-Mini S&P 500 futures as the trigger. Criminal charges were later brought against Navinder Sarao, a London-based trader, for "spoofing" activity contributing to the conditions that day.

The Intuition

The flash crash is the canonical example of how a modern electronic market can briefly disintegrate when liquidity providers step back. In a fragmented market with dozens of venues and many algorithmic market makers, liquidity is deep until it isn't. When the biggest providers pull their quotes simultaneously, the order book hollows out and prices can swing to absurd levels before human operators intervene.

How It Works

The SEC-CFTC report described the sequence as:

• Large sell program. A large mutual fund, Waddell & Reed, used an algorithm to sell 75,000 E-Mini S&P 500 futures contracts, worth roughly $4.1 billion. The algorithm targeted a participation rate based on volume, without price or time limits.
• Initial absorption by HFTs. High-frequency traders initially bought the futures, then quickly sold them to each other and to other market makers, generating "hot potato" volume that inflated the apparent participation rate and caused the algorithm to accelerate its selling.
• Liquidity withdrawal. As volatility rose, many HFTs and market makers widened spreads or pulled quotes entirely. Bid books thinned out.
• Cross-market spillover. Futures declines transmitted to cash equities and ETFs through index arbitrage. Stop-loss orders in individual stocks fired, hitting a market with almost no standing bids and producing "stub quotes" at prices like a penny.
• Recovery. A five-second pause in E-Mini futures trading, triggered by CME's "Stop Logic" functionality, let human traders reassess. Liquidity returned, prices rebounded, and trades at the worst prices were later broken.

Worked Example

Imagine a stop-loss order set at $40 on a stock that opened the day at $45. In normal markets, if the stock falls to $40 the stop triggers and the broker sells at the next available price, typically a few cents below $40.

On May 6 at 2:45 PM, that same stop hit in the middle of the liquidity void. With the bid book empty, the order crossed at the next available quote, which might have been a stub quote at one cent. The investor was filled at a price nearly 100% below where they intended. Trades more than 60% away from the pre-crash reference were later cancelled, but traders who sold at 30% declines had to live with those prints.

The practical implication was that stop-loss orders, widely used for risk management, could produce catastrophic fills in fragmented markets during stress. Exchanges subsequently introduced Limit Up-Limit Down (LULD) bands that halt individual stocks on extreme moves, and the SEC expanded circuit breakers.

Common Mistakes

• Blaming HFT alone. The joint report found that HFTs both provided and withdrew liquidity, amplifying the crash but not starting it. The trigger was a large institutional sell program with poorly calibrated parameters. HFT is a feature of the environment, not a single villain.
• Assuming the reforms solved it. LULD bands and updated circuit breakers make similar cascades harder but not impossible. Mini flash crashes in specific names and in Treasury markets (October 2014) have recurred.
• Using naked stop-loss orders in all conditions. A stop-loss without a limit can execute at absurd prices during liquidity events. Stop-limit or sell-limit orders constrain the worst fills but accept the risk that the order does not execute at all.
• Confusing depth with resilience. US equity markets show huge quoted liquidity in calm conditions. Depth is not resilience. It can evaporate in seconds when market makers pull quotes.
• Treating the event as purely technical. The flash crash illustrates a general principle: any automated system with feedback loops and risk-off rules can cascade. The same pattern appeared in the 1987 crash with portfolio insurance and in various bond-market stress events. The technology changes. The structural failure mode does not.

Frequently Asked Questions

Q: What was the 2010 flash crash in simple terms? On May 6, 2010, US equity markets fell roughly 9% in minutes and recovered almost fully within the same hour. A single large algorithmic sell order in futures sparked a chain reaction where high-frequency traders passed inventory to each other rapidly, market makers pulled their quotes, and individual stocks printed at absurd prices like one cent or $100,000.

Q: How does the 2010 flash crash affect investment decisions today? It shows that electronic market depth is shallow under stress. Investors who rely on stop-loss orders for risk management may receive catastrophic fills in a liquidity void. LULD bands added after 2010 help but have not prevented subsequent mini flash crashes in individual stocks and bond markets.

Q: What is a real-world example from the 2010 flash crash? Accenture traded at one cent per share for a brief moment during the crash, a stock that had been worth roughly $40 just minutes earlier. Stop-loss orders triggered by the first decline were filled at the cleared-out bid, executing at prices 99% below the pre-crash level. Exchanges later cancelled trades more than 60% from the reference price, but fills between 30–60% off stood.

Q: How can investors reduce flash-crash risk in their orders? Replace naked market stop-loss orders with stop-limit orders that specify a minimum acceptable fill price. Accept that this means the order may not execute in a fast market, but avoid the risk of a one-cent fill. Keep order parameters proportional to normal liquidity conditions.

Q: How is the 2010 flash crash different from the 1987 Black Monday crash? Both were driven by automated feedback loops, portfolio insurance in 1987 and algorithmic sell programs plus HFT hot-potato dynamics in 2010. The key difference is duration and depth: 1987 produced a 22.6% close-to-close loss that required months to recover, while the 2010 crash recovered almost fully within the same trading session.

Sources

1. SEC-CFTC. Findings Regarding the Market Events of May 6, 2010. https://www.sec.gov/news/studies/2010/marketevents-report.pdf
2. Kirilenko, A. et al. The Flash Crash: High Frequency Trading in an Electronic Market (CFTC). https://www.cftc.gov/sites/default/files/idc/groups/public/@economicanalysis/documents/file/oce_flashcrash0314.pdf
3. New York Fed Liberty Street Economics. The Flash Crash, Two Years On. https://libertystreeteconomics.newyorkfed.org/2012/05/the-flash-crash-two-years-on/

Disclaimer

This article is educational content only and is not financial advice. Nothing here is a recommendation to buy, sell, or hold any security. Consult a licensed advisor before making investment decisions.