The 2010 Flash Crash

What happened

The sequence, as established by the joint SEC-CFTC report and subsequent research:

Context. Markets were already stressed by European sovereign debt fears; volatility was elevated and liquidity thinner than normal.

The trigger. A large automated sell program in E-mini S&P 500 futures executed aggressively, using an algorithm targeting a percentage of volume without limits on price or time. In a thinning market, its own selling generated the volume that told it to sell faster.

The propagation. Futures pressure transmitted to the cash equity market through arbitrage. Liquidity providers, unable to interpret the disorder, widened quotes or withdrew. As displayed liquidity evaporated, incoming market orders walked down increasingly empty order books.

The absurd prints. With genuine bids gone, some orders executed against stub quotes — nominal placeholder quotes far from any sensible price that market makers post to satisfy obligations to quote at all. Hence trades at a penny, and at the other extreme, buy orders filling at ceiling prices.

The recovery. Within roughly twenty minutes, prices had substantially rebounded. Exchanges subsequently cancelled trades executed more than 60% away from pre-crash levels.

Why ETFs were disproportionately hit

Four reasons, and they interlock:

1. An ETF cannot be priced when its constituents cannot be. A market maker's quote comes from the basket's value (Chapter 32). When hundreds of underlying stocks were printing nonsense or not printing at all, the fair-value calculation broke. The rational response is to widen quotes drastically or step back, which is exactly what happened, and it was the correct behaviour rather than a failure of nerve.

2. Circuit breakers applied unevenly. The pre-2010 mechanism operated at the index level, halting the whole market when thresholds were breached, with individual securities largely unprotected. Nothing stopped a single ETF from trading at a penny.

3. Stop-loss orders became market orders in a void. This is the retail lesson (Chapter 15). Enormous numbers of resting stop orders triggered during the plunge, converted into market orders, and executed against whatever remained: often the stub quotes. Investors who believed they had protection discovered they had a mechanism for guaranteed selling at the worst possible price.

4. ETFs were where retail's automated orders lived. Diversified funds held by long-term investors carry more resting stop orders per dollar than individual stocks, so the mechanism above concentrated in them.

The cancellation statistic, 70% of broken trades in ETFs, measures all four, and specifically the fact that ETFs were where the market-structure gaps intersected with retail order types.

What it did not show

Three misreadings that persist and deserve direct rebuttal:

"ETFs caused the crash." The trigger was in futures, the propagation was market-structure-wide, and the SEC-CFTC report does not attribute causation to ETFs. They were where the damage became most visible.

"ETF arbitrage failed." It did what it structurally must: arbitrage requires a computable fair value, and for twenty minutes there was none. Once underlying prices became meaningful again, the mechanism restored ETF prices to NAV almost immediately: which is why the recovery in ETF pricing was as fast as the crash.

"ETF investors lost money." Holders who did nothing were unaffected; by the close, prices were near where they started. The losses were realised by those who sold during the window, overwhelmingly through stop orders and market orders. That distinction is the entire practical lesson.

The reforms it produced

The Flash Crash is the most consequential single event in modern US market-structure regulation, and much of what followed exists because of what happened to ETFs:

• Limit up/limit down (LULD). Individual securities now have dynamic price bands; trades outside the band cannot execute, and sustained pressure triggers a brief pause. A stock or ETF cannot legally print at a penny in normal circumstances.
• Revised market-wide circuit breakers, with lower thresholds and faster triggers.
• Elimination of stub quotes. Market makers must quote within a defined percentage of the national best bid and offer, so the placeholder prices that produced the worst fills are gone.
• Clearly erroneous trade rules were codified rather than improvised.
• Market-maker obligations were strengthened for registered participants.

Europe pursued parallel improvements: volatility interruption mechanisms already existed on several venues and were reinforced, and MiFID II later addressed algorithmic trading controls and market-making obligations directly.

Did it work? The 2015 test

On 24 August 2015, US markets opened sharply lower amid global volatility, and ETFs again experienced significant dislocations: many traded well below their NAVs in the opening minutes, with hundreds of trading halts under the new LULD mechanism.

The honest assessment:

• The extreme prints did not recur. No penny trades; the bands held.
• But dislocations still occurred, because the underlying cause was the same: an opening with hundreds of constituent stocks not yet trading meant fair value was again incomputable, and LULD halts on constituents made it worse in a feedback loop.
• The fix was further reform, not celebration: harmonisation of opening procedures, changes to LULD parameters, and continued work on the opening auction: because 2015 revealed that the machinery's weakest moment is the open (Chapter 12's rule for retail, derived from exactly this).

The conclusion for this book's purposes: the wrapper's vulnerability is not internal. It is inherited from the underlying market's ability to price itself, and every reform has been aimed at the underlying market's plumbing rather than at ETFs.

The lessons, by audience

For retail investors (the chapters this validates):

• Never use market orders in disorderly conditions (Chapter 15). Every catastrophic retail fill that day was a market order or a triggered stop.
• Stop-loss orders on diversified holdings are a mechanism for realising losses at the worst prices (Chapter 15's argued position, evidenced here).
• Avoid the open (Chapter 12). Both 2010 and 2015 concentrated their damage where price discovery was weakest.
• Doing nothing was the winning strategy, as it usually is in events measured in minutes.

For institutions: limit-based execution and RFQ channels (Chapter 72) function when screens do not; fair-value estimation independent of the tape is what allows trading through disorder.

For the structural debate: ETFs are a magnifying glass on underlying market quality. When the underlying prices well, ETFs price well; when it does not, the ETF displays the failure in a single visible number, which is a service that looks like a scandal.

A worked example

Two investors held the same broad US equity ETF on 6 May 2010, each with $50,000.

• Investor A had a stop-loss order 10% below market, placed months earlier "for protection." At 14:45 the stop triggered into a market with almost no bids and filled at a price roughly 38% below the pre-crash level. By the close, the ETF had recovered to near its opening price. Realised loss: about $19,000, permanently, from a market event that lasted twenty minutes.
• Investor B had no conditional orders and was not watching. Her account showed a normal day's small loss at the close, and she learned about the Flash Crash from the news.

The market did the same thing to both of them. The instrument that determined the outcome was an order type, chosen in advance, in calm conditions, by someone who believed it reduced risk.