A flash crash is a rapid, deep fall in prices followed by a nearly equal recovery, and the episode argues the cause is usually not aggressive selling but a sudden withdrawal of liquidity as traders and algorithms pull their orders. On May 6, 2010, the Dow fell over 1,000 points and recovered most of it in about 15 minutes. The U.S. Department of Justice blamed Navinder Singh Sarao, a trader working largely from a suburban house, and charged him with 22 criminal counts, including fraud and market manipulation, over spoofing: placing huge sell orders and canceling them milliseconds before execution. On June 22, 2017, a legitimate multimillion-dollar sale on the GDAX exchange pushed Ethereum from about $317 to $224, triggering roughly 800 stop-loss and margin orders that drove it to 10 cents.

On May 2, 2022, a Citigroup trader in London made an input error that erased about 300 billion euros across European markets in five minutes, with Sweden down 6.8 percent, Denmark 6.7 percent and Finland 7.5 percent, and Nasdaq said the trades would stand. On October 7, 2016, the British pound fell over 6 percent in two minutes to its lowest against the dollar since 1985, possibly because news-reading algorithms reacted to hard Brexit headlines. Singapore’s October 2013 crash erased $6.9 billion, and in August 2014 the exchange introduced a 20-cent minimum price, short position reporting and a 5 percent collateral levy, yet mini-crashes returned that November.

• Liquidity in modern markets is described as a mile wide but only an inch deep.
• A falling knife, unlike a flash crash, is a drop on real bad news that stays down.
• On January 2, 2019, the yen and Australian dollar moved over 4 percent in minutes, with an Apple forecast blamed even though it came an hour earlier.
• Some Singapore stocks lost up to 87 percent of their value in the 2013 crash.
• In November 2014, 40 stocks moved more than 1 percent in under a second, and another 88 did the same.