What is a Flash Crash? Definition, Formula, and Example
A flash crash is a rapid, deep, and largely self-reversing price collapse across a market or asset that unfolds in minutes rather than sessions, typically driven by liquidity withdrawal and algorithmic feedback loops rather than fundamentals.
What is a flash crash?
A flash crash is a severe price drop of an index, sector, or single security that happens over minutes — sometimes seconds — and then substantially reverses within the same trading session. The defining feature isn't just magnitude, it's speed combined with a lack of fundamental trigger: no earnings miss, no macro data surprise, no news headline explains the move. What actually happens is a liquidity vacuum. Market makers and high-frequency trading (HFT) firms that normally provide continuous two-sided quotes pull their bids simultaneously — either because risk limits trip, volatility models flag danger, or a large order overwhelms the order book — and price discovery breaks down. The next resting bid might be 5-10% below the last trade instead of a fraction of a percent, and market orders slam through that gap.
How a flash crash is identified
There's no single regulatory formula, but exchanges and researchers generally classify an event as a flash crash when it meets three conditions: (1) a decline of several percent in an index or a much larger percentage move in an individual name, (2) completed in under 15-20 minutes, and (3) at least a partial recovery (commonly 50%+ of the drop) within hours. The canonical case is May 6, 2010: the Dow Jones Industrial Average fell about 998 points (~9%) between 2:32 PM and 2:45 PM ET, then recovered most of the loss by 3:07 PM. The SEC/CFTC joint report attributed the trigger to a $4.1 billion E-mini S&P 500 futures sell program executed by a single trader (Waddell & Reed) without regard to price or time, which HFT firms initially absorbed and then aggressively resold, draining liquidity as they hit their own risk limits.
Worked example
On May 6, 2010, PG traded from roughly $60 down to $39.37 — a 35% intraday drop — before closing near its pre-crash level. Accenture (ACN) printed a trade at $0.01. These weren't rational sellers valuing the companies at pennies; they were market orders executing against a nearly empty book after market makers stepped away and stub quotes (placeholder bids like a penny, meant never to be hit) became the only liquidity left. The exchanges canceled trades that moved more than 60% from the pre-crash price under "clearly erroneous" rules, but thousands of retail stop-loss orders had already filled at absurd prices before the busts were processed.
When traders watch for flash crash risk
Retail and professional traders alike use wide, mental stops or stop-limit orders instead of plain market stop-loss orders in thin or after-hours conditions, precisely because a stop-loss order can fill at a flash-crash price far below the trigger. Options market makers watch for flash-crash-style dislocations because they blow out implied volatility and bid-ask spreads instantaneously, and algo desks build "kill switches" that halt trading when price moves exceed a volatility-adjusted threshold in a short window. Post-2010, exchanges layered in single-stock circuit breakers and, later, the limit up-limit down (LULD) mechanism specifically to slow this kind of cascade before it reaches 2010-level severity.
Limitations and misconceptions
A flash crash is not proof of manipulation by itself — the 2010 event was originally blamed partly on spoofing by trader Navinder Sarao, but the primary SEC/CFTC finding was structural (liquidity withdrawal cascading through HFT algos), not a single bad actor. It's also not unique to equities: the October 15, 2014 Treasury flash rally, the August 2015 ETF-pricing dislocation, and the January 2019 USD/JPY flash crash all show the same liquidity-vacuum mechanics in bonds, ETPs, and FX. Don't assume a flash crash means the underlying business or macro picture changed — the defining feature is that fundamentals didn't move, only the order book did.