On Thursday, the phones at Goldman Sachs were about to explode.
"The level of pain is extremely high."
That was the sentence Goldman Sachs' top equity trader John Flood sent to clients in his Thursday daily report. His phone, in his own words, was "exploding."
The Nasdaq 100 index closed down 6.4% on Thursday, shedding approximately $2.3 trillion in market value in a single day — the largest one-day decline since March 2020. The S&P 500 fell 4.4%, and the Philadelphia Semiconductor Index plunged 12%. The "fear index" VIX surged 70%, Nvidia gave back all of its year-to-date gains, and SanDisk plummeted 22% in a single day.
The direct trigger for this crash was the Democratic Party's sweeping victory in both chambers of Congress in the midterm elections. Goldman Sachs immediately warned that this "blue tsunami" would put approximately 45% of planned US data center expansion capacity at risk of being paused — the core narrative supporting AI tech stock valuations was suddenly called into question overnight by political reality.
But the ferocity of the market decline far exceeded what the policy shock alone could explain.
Nomura derivatives strategist Charlie McElligott, in a report published intraday on Thursday, offered another layer of explanation — a considerable portion of this crash was caused by "the smartest people in the room" stepping on themselves.
McElligott described a market paradox: during Thursday's US trading session, after Morgan Stanley calculated $52 billion in mechanical sell orders before the close at midday and all of Wall Street knew about it, everyone tried to sell early and buy back at the lows — but the front-runners themselves became the force driving the decline. He warned that if the usual "2-3 PM low then bounce" failed to materialize, the front-runners themselves would become the force driving the market down. Sure enough, the close was brutal. Combined with $47 billion from leveraged ETFs, $61 billion from volatility control funds, and dealer negative gamma, total mechanical selling pressure exceeded $100 billion — and it would persist into Friday.
And today in Asia-Pacific markets, the sharp declines in South Korea and Japan are answering that question.
A $100 Billion Time Bomb That All of Wall Street Knew About
To understand McElligott's logic, one must first understand a mechanism.
There is a type of product in the market called leveraged ETFs, such as "2x long Nasdaq." They must do one thing before the close every day: rebalance the day's gains and losses back to their target leverage ratio. When the market drops sharply, this means they must sell large amounts of stock before the close — the worse the decline, the more they sell. This is passive, mechanical, and unavoidable.
There is another category called volatility control funds, which systematically reduce equity positions when market volatility spikes — equally mechanical, and typically executed with a one-to-two-day lag.
On Thursday, these two forces combined, and the scale was terrifying.
Morgan Stanley's derivatives team calculated at midday on November 5 local time: assuming prices held at the 1 PM level, forced selling from leveraged ETFs before the close would be approximately $52 billion. The semiconductor sector alone would account for $21 billion, with the Nasdaq 100 absorbing another $15 billion.
Nomura's McElligott arrived at a slightly lower but equally alarming figure: approximately $47 billion from leveraged ETFs — the second-largest single-day rebalancing in Nomura's database history, behind only $51.3 billion on June 5 this year. He added that this sell order was "especially painful" because 86% of the assets were concentrated in tech and AI-related names.
Combined with approximately $61 billion in sell orders from volatility control funds, plus roughly $7 billion in selling pressure from dealer negative gamma for every 1% decline, total mechanical supply across the market easily exceeded $100 billion.
The key characteristic of these numbers is that they could be calculated in advance.
After Morgan Stanley calculated the $52 billion figure at midday, in McElligott's words, "on Wall Street, no piece of information stays secret for more than 45 minutes" — the number quickly became "known across all of Wall Street."
The Smart Money's Clever Plan: Sell Early, Wait for the Low, Buy Back
Knowing about this mechanical sell order, what was the traders' reaction?
McElligott said that active traders facing this kind of mechanical flow typically try to "front-run" it.
The logic is simple: since there will be a wave of large-scale passive selling before the close, I'll sell now and buy back at a lower price after it's done.
And this strategy had been working recently. McElligott noted in his report that in several recent "single-day sharp swing" trading sessions, the market tended to form its daily low between 2 PM and 3 PM Eastern Time, after which front-runners and hedgers began to reverse, take profits, and close positions, pushing the market to bounce before the close — with the final closing price ending up higher than the extreme implied by the "expected" mechanical selling.
Just last Thursday, McElligott had reminded the market that "it's happening again" — at that time, leveraged ETFs faced $35 billion in sell orders, and the market ultimately followed the classic structure of a "2-3 PM low plus a closing bounce."
During the Korean market crash in June this year, US stocks followed a similar script.
On the day the tariff pause was announced in April, the mechanism worked in reverse: McElligott calculated the largest-ever +$34.3 billion pre-close buy rebalancing, front-runners bought early, and ultimately drove the market sharply higher.
So on Thursday, when Morgan Stanley's $52 billion figure spread across the Street at midday, every trader who knew the number was thinking the same thing: sell now, wait for the 2-3 PM low, then buy back.
The Reflexivity Trap: The Cavalry Is the One Selling
Then McElligott wrote down that chilling concern in his report: "What if that 'assumed 2-3 PM low then bounce' simply doesn't materialize?"
Obviously, the scary part now is this: these numbers are so large that 'what if' we don't get that almost-taken-for-granted late-session 'low first then bounce' (a standard one-to-two percent bounce from the low), unable to bounce during the 2-3 PM 'standard TWAP' window and instead making new lows... but rather, after feinting at the market's assumption that 'it will bounce in the final hour,' closing at the lowest point.
This is precisely the core of the reflexivity paradox.
When everyone knows the same script and acts on it simultaneously, the script itself becomes invalid.
The $100 billion in mechanical sell orders that should have been concentrated before the close were instead spread throughout the entire day by active traders "front-running" — and these front-runners themselves became the force driving the market down.
At the same time, volatility control funds' sell orders had a one-to-two-day lag — Thursday's crash would trigger these funds to continue selling on Friday. McElligott wrote that the volatility control selling volume corresponding to the decline at 3 PM that day was approximately $61 billion, and if the decline widened to 10%, that figure would swell to $70 billion.
More critically: when all the front-runners have already sold and are waiting for "that low" — who will be there to buy?
McElligott used a meme in his report to describe the scene: everyone is sitting in chairs, waiting for the same bounce — but there aren't enough chairs.
A ZeroHedge analysis summarized it this way:
"Sometimes the cavalry arrives at 3 PM. Sometimes, the cavalry is the one selling."
On Thursday, the cavalry didn't come. The market closed at its lowest point of the day: Nasdaq 100 down 6.4%, S&P 500 down 4.4%.
The Stampede Scene: The Trade of the Year Gets Trampled in 48 Hours
Before the reflexivity mechanism detonated, there was another powder keg waiting for a fuse.
McElligott called this year's "trade of the year" on Wall Street the "semiconductor-energy barbell strategy" — long semiconductors (benefiting from AI computing demand) while simultaneously long energy (geopolitical risk premium plus inflation hedge). This strategy had accumulated returns of 66% just a few weeks ago, with a Sharpe ratio of 3.3.
Then, over the past 48 hours, both legs were taken out simultaneously:
Energy: down 10.2% this week
Semiconductors: down 13.1% this week
Weekly Sharpe ratios were -17.2 and -10.1 respectively.
The collapse of the energy leg came from the market repricing the Iran situation — McElligott observed that the White House signaled it might "soften" its stance on Iran, the war premium was rapidly drained, and Brent crude posted its largest weekly decline since the COVID-19 pandemic.
The collapse of the semiconductor leg came from market concerns about data center regulation after the Democratic Party's sweep of Congress.
With both directions collapsing simultaneously, funds holding the barbell strategy faced not just losses, but forced liquidation. The mentality McElligott described of "preserving this year's gains and surviving into next year" was completely shattered at this moment.
Walleye Capital announced that day it would return client funds and fired its head of equities, calling it the worst month since COVID-19. Goldman Sachs trader John Flood described long-only institutions as "once again deadlocked," with hyperscaler positions bought during yesterday's bounce being rapidly dumped, and "no buyers stepping in to catch the falling knife yet."
Abnormal ETF trading volumes confirmed the level of panic: SPY volume was 160% above its 20-day average, QQQ was 100% higher, and semiconductor ETF SMH was 150% higher.
According to Vanda Research senior equity strategist Jai Malhi, Thursday's retail investor selling volume was "on track to be the largest retail liquidation since late July last year."
Bonds Get Squeezed, Another Leg Is Also Bleeding
There was another overlooked leg breaking at the same time.
CTA trend funds had accumulated large short bond positions this year as part of the "short bonds/long inflation" trade — an implicit companion to the barbell strategy.
The stock market crash triggered a safe-haven rush into bonds, US Treasury prices surged, and bond shorts were forced to cover. McElligott's calculations showed that CTAs faced a "massive short squeeze" in the bond direction: if US Treasury prices rose another 2%, it would trigger approximately $225 billion in forced global bond buying, including about $48.5 billion in US 2-year bonds, about $69.5 billion in Eurozone 2-year bonds, and about $23.9 billion in Japanese 10-year bonds.
This means the bond short squeeze itself was becoming a force driving bond prices higher and further compressing equity longs — yet another reflexivity loop.
The Crash Isn't Over: Friday's Lagged Selling Pressure
Thursday's close did not mean the end of this round of mechanical selling pressure.
Volatility control fund execution had a lag. McElligott wrote that Thursday's market decline would continue to trigger sell orders from these funds on Friday — "today's decline is loading further into the queue."
UBS's estimate was more direct: its risk control rebalancing model showed that if the S&P 500 fell 2%, systematic fund selling would reach $95 billion to $105 billion.
Asian markets were already answering the question.
On Friday morning, South Korea's KOSPI plunged 8.8%, breaking below the 5,000-point mark and triggering a circuit breaker — a level that South Korean President Lee Jae-myung had once viewed as a milestone for economic recovery, now rapidly breached. Japan's Nikkei 225 fell nearly 5.8%, breaking below 60,000 intraday and dropping more than 20% from its June high, officially entering a technical bear market.
US stock futures saw a modest bounce in early Asian trading, with Nasdaq futures briefly turning positive, suggesting some investors were starting to buy the dip. But UBS warned that the pressure from systematic selling had not yet fully released.
There is one more variable on Friday: the US October nonfarm payrolls report will be released that day, with market consensus expecting 59,000 new jobs and the unemployment rate holding at 4.2%. The direction of the data will determine whether this round of adjustment can find a window for breathing room.