Key Takeaways
- • Jordi Visser, president and chief strategist of 22V Research, argues the November 2026 AI-stock selloff was a forced hedge fund deleveraging event, not the popping of an AI bubble, citing JPMorgan's estimate that trend-following funds and levered ETFs cut more than $70 billion of equity exposure within 90 minutes of the Thursday open.
- • Goldman Sachs' panic gauge reached 9.57 out of 10, producing the largest two-day swing on record, and Morgan Stanley recorded a 5% single-day drop in US long-short net leverage, the fifth largest since 2010.
- • For the first time since tracking began in 2016, both major hedge fund cohorts lost more than 3% in the same month, according to the primary source's account.
- • Visser has been buying the decline for three weeks, most heavily Micron and Marvell, anchoring entries to the 200-day moving average, while stating that the deleveraging is "just the beginning" and that a bottom is not yet called.
- • Visser's core framework holds that the S&P 500 has become "492 bonds and 8 call options," making factor flows rather than index-level fundamentals the decisive driver of prices.
Summary
Jordi Visser, president and chief strategist of 22V Research, speaking on his own podcast, characterized the violent November 2026 selloff in AI stocks as a hedge fund deleveraging event rather than the bursting of an AI bubble. He cites JPMorgan's estimate that trend-following funds and levered ETFs dumped more than $70 billion of equity exposure within 90 minutes of the Thursday, November 5 open. Net selling of S&P futures reached $11 billion that day, CTAs flipped to net short, and levered ETFs saw the largest amount to sell in data going back to 2007, per Visser. "The AI bubble didn't pop. Hedge fund leverage did," he said.
The market signals cited were extreme. Goldman Sachs' panic gauge hit 9.57 out of 10, the largest two-day swing on record. Morgan Stanley clocked a 5% single-day decline in US long-short net leverage, the fifth largest since 2010. The VIX's one-day percentage move ranked third in 25 years, and the VVIX posted its biggest one-day jump in 20 years. Equal-weight S&P fell about 1% for the week, while QQQ and SMH fell 5% and 11%, respectively; SMH remained up 29% year to date. Visser also pointed to AQR's publicly visible mutual fund returns, where the long-short fund had its worst seven-day reading outside COVID, and said "This feels like the 2007 quant quake."
Visser does not call a bottom. He expects funds to hedge up net exposure but to be unable to reduce gross exposure fast enough, which he identifies as the binding constraint this time. He notes that long-short funds are underweight AI for the first time since 2023, but says this is relative to a benchmark in which AI now comprises more than half the S&P 500. He also argues that investor redemptions have not fully shown up and that "much money will be handed back even without formal redemptions."
Despite these views, Visser has been buying for three weeks, beginning around October 27, 2026, with his largest single week of buying during the recording week. He bought Micron most heavily, plus Marvell and Nvidia, and sold other optical names at a loss for tax purposes. He anchors valuation to the 200-day moving average: Micron moved from 200% above that average to 13% below as of Friday's close. He concedes Micron could fall toward $500 before bottoming. Visser draws an analogy to software, where IGV's breakdown was followed by recovery within about three months despite stable earnings.
Visser's structural thesis is that the S&P 500 "is no longer the S&P 500," but rather "492 bonds and 8 call options," with the 492 non-AI companies trading as bonds between roughly 70 and 95 cents on the dollar. He also warns that the bear case remains live: if deleveraging unwinds quickly rather than bleeding out, the market could fall much lower, and he acknowledges the fundamental long-short model may be structurally broken.
The market signals cited were extreme. Goldman Sachs' panic gauge hit 9.57 out of 10, the largest two-day swing on record. Morgan Stanley clocked a 5% single-day decline in US long-short net leverage, the fifth largest since 2010. The VIX's one-day percentage move ranked third in 25 years, and the VVIX posted its biggest one-day jump in 20 years. Equal-weight S&P fell about 1% for the week, while QQQ and SMH fell 5% and 11%, respectively; SMH remained up 29% year to date. Visser also pointed to AQR's publicly visible mutual fund returns, where the long-short fund had its worst seven-day reading outside COVID, and said "This feels like the 2007 quant quake."
Visser does not call a bottom. He expects funds to hedge up net exposure but to be unable to reduce gross exposure fast enough, which he identifies as the binding constraint this time. He notes that long-short funds are underweight AI for the first time since 2023, but says this is relative to a benchmark in which AI now comprises more than half the S&P 500. He also argues that investor redemptions have not fully shown up and that "much money will be handed back even without formal redemptions."
Despite these views, Visser has been buying for three weeks, beginning around October 27, 2026, with his largest single week of buying during the recording week. He bought Micron most heavily, plus Marvell and Nvidia, and sold other optical names at a loss for tax purposes. He anchors valuation to the 200-day moving average: Micron moved from 200% above that average to 13% below as of Friday's close. He concedes Micron could fall toward $500 before bottoming. Visser draws an analogy to software, where IGV's breakdown was followed by recovery within about three months despite stable earnings.
Visser's structural thesis is that the S&P 500 "is no longer the S&P 500," but rather "492 bonds and 8 call options," with the 492 non-AI companies trading as bonds between roughly 70 and 95 cents on the dollar. He also warns that the bear case remains live: if deleveraging unwinds quickly rather than bleeding out, the market could fall much lower, and he acknowledges the fundamental long-short model may be structurally broken.
Go deeper than the headline
Research every firm in this story, track the ones that matter and get the daily briefing — free for 7 days.