When will deleveraging in tech stocks end? Institutions: Key to watch two major signals

When will deleveraging in tech stocks end? Institutions: Key to watch two major signals

```

Deleveraging on the trading end is currently driving recent fluctuations in the A-share technology sector, but the key signals needed to confirm a bottom have not yet appeared.

Guojin Securities' strategy noted in its July 12 report that the current transaction volume of the TMT sector as a proportion of the entire A-share market remains high, and the concentration of trading in popular sectors has not declined significantly, nor has bullish sentiment reverted to normal levels. Referring to previous deleveraging experiences, the bottom for this round of adjustment is yet to be confirmed, and trading-driven volatility may persist for some time.

Two prerequisites must be met for trading-driven volatility to bottom out: First, the concentration of trading in formerly popular sectors declines significantly; second, investors’ bullish sentiment returns to normal levels.

Triple signals confirm deleveraging

Last week, the STAR 50 Index saw a maximum pullback of over 5%, and A-share margin balances decreased by more than 30 billion yuan in a single week, with the activity of margin buying dropping to an annual low. Meanwhile, the technology sectors in both Japan and South Korea saw marked adjustments, with margin balances in both places consistently falling since July, indicating a degree of linkage in deleveraging on the trading end of Asian tech stocks.

Furthermore, in most cases, the implied volatility in options pricing should be higher than the realized volatility in the market. However, currently, the implied volatility for STAR 50 Index options and the one-month implied volatility for KOSPI 200 options are both lower than the realized volatility from the past 20 trading days; that is, option investors, despite already elevated volatility, are unwilling to pay an even higher volatility premium. Since options trading itself is highly leveraged, the negative volatility premium further confirms the deleveraging on the trading end.

Guojin Securities characterizes deleveraging in A-share trading from three dimensions: declining or sharply slowed growth in margin balances, rapid decrease in margin buying activity, and implied volatility in the options market falling below realized volatility.

All three indicators have now been triggered. Last week, A-share margin balances decreased by more than 30 billion yuan; margin buying as a proportion of total A-share turnover fell to a yearly low. The implied volatility for STAR 50 Index options has dropped below the realized volatility of the past 20 trading days, and similarly, the one-month implied volatility for KOSPI 200 index options is below the 30-day realized volatility.

This anomaly in the options market is particularly noteworthy. Normally, implied volatility in options pricing should be higher than realized volatility, but at present, option investors are unwilling to pay a higher volatility premium despite already high volatility levels. Given the high leverage inherent in option trading, a negative volatility premium is itself evidence of trading-end deleveraging.

Industry-level data also supports this logic. Over the past month, the industry with the highest proportion of net margin buying relative to total turnover was communications and electronics, and these two sectors were also the most volatile on the A-share market last week.

Bottom confirmation awaits two conditions

Since the launch of the rally on September 24, 2024, A-shares have undergone four rounds of phased trading-end deleveraging, taking place in November 2024, March 2025, October to November 2025, and mid-June 2026, each corresponding to periods of increased volatility after market surges.

Reviewing the past four experiences, two prerequisites must be met for trading-driven volatility to bottom out: First, the concentration of trading in formerly popular sectors declines significantly, specifically shown as TMT sector (especially electronics) turnover as a proportion of total A-share turnover recedes; second, investors’ bullish sentiment returns to normal, with noticeable narrowing in sentiment indicators such as implied volatility for technology indexes like STAR 50, and the spread between implied volatility for call and put options.

Guojin Securities points out that the volatility premium can also serve as a monitoring tool. Currently, the five-day moving average volatility premium for STAR 50 Index is -2.2%, while in previous trading-driven volatility periods, the lowest value of this indicator was about -8%. The end of trading-driven volatility usually accompanies a return of the volatility premium to normal levels. At present, neither of these two conditions has been met; the overall TMT sector turnover proportion remains at a high level.

Trading-driven volatility is not equivalent to a trend-based decline

Historical experience shows that trading-end deleveraging-induced volatility is often just the driver for the market's first wave of decline, while the direction of fundamentals is the core variable determining whether the market enters a sustained downward phase.

In June 2015, as regulatory scrutiny of off-balance sheet financing increased, A-share margin balances quickly fell, and trading factors drove the first pullback of the mobile internet index, followed by a period of volatile rebound. However, starting from Q4 2016, revenue growth, gross margin, and ROE of the mobile internet sector all continuously declined, and weakening fundamentals eventually pushed the index into a prolonged downturn.

The first wave of decline after the Ning combination peaked in November 2021 similarly came with a marked decrease in margin balances. Since fundamentals had not comprehensively deteriorated at that time, the Ning combination rebounded in Q2 2022; but from Q4 2022 onward, revenue and net profit growth both weakened, and the index entered a clear downward phase.

For the current AI-related technology sector, fundamentals remain relatively steady. Recently, both the token usage and weighted market prices for large models have rebounded; last week, Meta announced an investment of C$13 billion in Canada to build a data center—somewhat alleviating market concerns about excess computing power. According to Guojin Securities’ technology cycle framework, AI physical consumption index is still rising, and a turning point for this indicator will be the key basis for judging whether the next wave of sustained decline emerges.

Risk warning and disclaimerThe market has risks, and investment requires caution. This article does not constitute personal investment advice and does not take into account the specific investment objectives, financial situation, or needs of individual users. Users should assess whether any opinions, viewpoints, or conclusions in this article are appropriate for their own circumstances. Investment is at your own risk. ```