Key Takeaways
A major deleveraging wave is sweeping across the US ETF market. The notional value of the 200 largest US ETFs — a measure that captures total market exposure including the effect of leverage — has declined by -$100 billion, or -20%, over the last month, falling to approximately $400 billion, the lowest level since May. To illustrate the concept, a leveraged ETF with $100 billion in assets and 2x leverage would provide $200 billion in notional exposure; when that exposure contracts sharply, it signals either falling asset values, leverage reduction, or both occurring simultaneously.
This decline follows an even larger expansion: notional exposure surged by +$200 billion, or +67%, since the March 30th market bottom, meaning the market added leverage aggressively during the recovery rally before this month's reversal began. For context, during the prior deleveraging event from the November 2024 peak to the April 2025 bottom, notional value fell by -$130 billion, or -37% — a steeper percentage decline than what has occurred so far this cycle. Despite the current pullback, leveraged ETF exposure remains more than double (+100%) the peak level observed during the 2021 meme-stock frenzy, underscoring how much structural leverage has built up in the ETF ecosystem over the past several years.
QQQ remains in a downtrend despite a recent rally session. The middle channel support band sits between 535.00 and 540.00, while resistance is located between 720.00 and 725.00. The MACD indicator showed strong momentum during the rally, meaning another leg higher toward resistance cannot be ruled out. However, QQQ is not currently in an oversold position, which would be required to generate a fresh intermediate-term buy signal under standard technical criteria. This combination — downtrend structure with a momentum-driven bounce but no oversold confirmation — typically points to a corrective rally within a larger downtrend rather than a trend reversal.
Retail investors dumped stocks yesterday at the fastest pace since the COVID-19 crash, a signal that typically accompanies capitulation phases rather than measured de-risking. This behavior is playing out on a much larger scale in South Korea, where the KOSPI has plunged more than 33% this month. That decline is on track to become the index's worst month in history, surpassing both the October 1997 IMF Crisis (-27%) and the October 2008 Global Financial Crisis (-23%).
The mechanism behind the collapse is heavily leverage-driven. Goldman Sachs estimates that more than 1.2 million leveraged retail trading accounts in South Korea triggered margin calls as of July 13th, with an estimated 320,000 to 360,000 accounts fully liquidated. That means roughly 3.4% of South Korea's entire adult population has received a margin call this month alone. The KOSPI has fallen an additional ~18% since July 13th, and the total number of fully liquidated accounts is now believed to exceed 500,000 — indicating the forced-selling cascade has continued to accelerate even after the initial wave of margin calls.
The current US ETF deleveraging and South Korea's margin-call cascade both echo dynamics from the 2000-2002 Dot-Com bust, when excessive retail leverage on richly valued technology stocks unwound over roughly two years, ultimately dragging the Nasdaq down approximately 78% from peak to trough. Two key differences stand out. First, the speed of the current unwind — a 20% drop in ETF notional value in a single month, and a 33% KOSPI decline in the same period — is compressed into a dramatically shorter timeframe than the multi-year Dot-Com decline. Second, the current cycle is occurring against a backdrop of substantially higher absolute leverage levels, given that leveraged ETF exposure remains double the 2021 meme-stock peak even after this month's pullback. This suggests the mechanics of forced selling — margin calls triggering further price declines that trigger more margin calls — are structurally similar to 2000-2002, but the leverage embedded in modern ETF products may allow the unwind to move faster and with greater intraday volatility than the original Dot-Com deleveraging.
Based on the leveraged ETF flow data and the concentration of retail margin activity, leverage accumulation appears heavily skewed toward high-beta growth sectors:
Most leverage accumulated:
Least leverage accumulated:
Given the deleveraging backdrop, a rotation away from the most leverage-saturated NASDAQ segments toward lower-beta, less leveraged names is consistent with historical post-deleveraging behavior. This does not necessarily mean abandoning growth exposure entirely, but rather favoring balance-sheet-strong, profitable technology names over speculative, leverage-favored momentum names until the ETF notional contraction stabilizes.
Tickeron's AI Trading Bots incorporate sector-level signals to identify where capital is rotating in real time, cross-referencing price action, volume, and leverage-flow proxies across ETF categories. The Financial Learning Models (FLM) component analyzes individual stock and ETF trend structures — including moving average convergence, momentum divergence, and support/resistance zones like those currently defining QQQ's channel — to generate directional forecasts that account for both the broader sector rotation and the individual security's technical trend state. In periods of rapid deleveraging such as the current one, the FLM engine is designed to flag deteriorating trend structures earlier than lagging indicators, which is particularly relevant given the compressed timeline of this unwind relative to historical deleveraging cycles like the Dot-Com bust.
Tickeron AI Perspective