10 Things to Watch Most During the Unexpected Pullback | Don't Make Decisions in Panic

Over the past two weeks, China's A-share tech sector has experienced a "rapid plunge" that left everyone gasping for air.

As of the close on July 17, the STAR 50 Index plummeted 16.93% in a single week, marking the largest weekly decline since the STAR Market's inception seven years ago; year-to-date drawdowns since July have exceeded 22%. The ChiNext Index fared no better, dropping 10.78% in a single week with cumulative monthly losses of approximately 21%. The situation was even more dire in specific sub-sectors: the electronics and communications sectors fell by approximately 31% and 24%, respectively, while previously hot areas such as computing power, storage, and semiconductors were almost completely wiped out. (Data source: Sina Finance "Another Wave of A-Share Deleveraging", International Financial News "How Long Will Tech Stocks Fall?")

The more extreme the situation, the easier it is to be led by emotions.

Today, we will thoroughly address the 10 questions that concern everyone most.No chicken soup, no slogans—just data and logic to help you see through the essence of this adjustment and determine exactly what to do next.

Q1: Why is tech falling so sharply? Has the fundamental outlook collapsed?

This sell-off is essentially the result of three overlapping factors resonating together, largely unrelated to whether the AI industry itself is healthy:

First, global tech trading is undergoing "deleveraging." Overseas markets crashed first, dragging us down. The Korean market was the "epicenter" of this leverage buildup, with single-stock leveraged ETF scales briefly exceeding 16 trillion KRW. When the Bank of Korea unexpectedly raised interest rates combined with deleveraging measures in mid-July, leveraged funds fled en masse. The KOSPI retreated nearly 25% from its highs, and semiconductor leaders like Samsung and SK Hynix plummeted. The same occurred in US stocks, where leveraged semiconductor ETF scales shrank by $63 billion from their June peak. Global AI asset trading logic is highly intertwined; when overseas markets shake, risk appetite for A-share computing, storage, and chip sectors follows suit.

Second, trading positions are too crowded, leading to concentrated profit-taking. Before the adjustment, the top 5% of A-share turnover accounted for a record high, placing crowding levels at historical extremes. What does this mean? Everyone is crammed onto the same tech boat. If the boat rocks slightly, everyone rushes for the exit, causing a stampede. Additionally, with the interim report disclosure window approaching, the market had already priced in earnings expectations for AI computing. At the realization node, capital naturally chose to "secure profits."

Third, emotions amplified the declines. Margin accounts, quantitative strategies, and take-profit orders sold simultaneously, creating a negative feedback loop of "fall → trigger liquidation → further fall." This is a typical clearing of trading structures, which is a separate issue from whether companies are profitable or if the industry is viable.

In short:The industry hasn't broken; the chips are just scrambled. Once the crowded trades digest, the capital that should return will come back.

Q2: When will the impact from overseas ease?

This volatility in the Korean market is akin to a "leverage bubble burst." While fundamentals aren't fundamentally flawed, the splashing water certainly affects A-shares.

To judge whether the overseas shock is subsiding, focus on tracking changes in four dimensions:

① Whether leveraged funds have been fully cleared. The most direct indicator is when the scale of Korean leveraged ETFs stops shrinking rapidly and when margin balances decline consecutively. Currently, margin balances have only dropped nearly 8% from their peaks, so it cannot be said they are fully cleared yet.

② Whether client margins have stabilized. If margins continue to drain but margin balances don't drop, it means investors are still passively holding, and risks haven't been fully released. True stabilization should see "margin stop-loss + continued decline in margin balances" occur simultaneously.

③ Whether semiconductor fundamentals can take over. Korean stocks can only stabilize if heavyweight stocks like Samsung and SK Hynix hold firm. Next, capital expenditure guidance from overseas cloud providers, price trends for HBM and DRAM, and financial report results will determine whether the market can switch from "liquidity panic" back to "earnings pricing."

④ Whether regulators intervene to cut off the negative feedback. For example, restricting new leverage, raising entry thresholds, or injecting stable capital. However, if there is only market support without deleveraging, a second round of clearing may follow after the rebound.

Overall,the extreme volatility caused by overseas deleveraging will gradually ease in the future.

Q3: Where are we in this tech adjustment? How much longer will it fall?

First, clarify this:A market bottom is never a single point, but a process of repeated grinding. What we can do is learn from history, roughly judge if the adjustment is sufficient based on past performance, and then use real-world signals to confirm if risks are easing.

Historical data shows that intermediate adjustments in A-share main-line rallies last an average of about 20 trading days, with average drawdowns around 18%; median adjustment durations for pan-tech main lines are about 17 trading days, with drawdowns of about 16%. If it's an industry-level "bull market pullback," the adjustment cycle is usually 2-3 months, with magnitudes ranging from 20%-35%. (Data source: GF Securities Development Research Center "Adjustment Duration and Magnitude of Main Lines Since 2012")

How to view the current position?: In terms of space, the STAR 50 has drawn down more than 20% from its highs, and some upstream computing tracks have seen maximum drawdowns of 30%-40%, meaning significant risk has already been released. In terms of time, the adjustment started only about two weeks ago. Chip turnover, valuation digestion, and earnings verification all take time, so there will likely be a process of repeated bottom-grinding ahead.

To judge if it's nearing the end, observe these five types of signals:

• Trading volume and volatility retreat from extreme levels, with no more panic dives at the close

• Margin balances and leveraged product scales continue to shrink, and market crowding clearly cools down

• Sector leaders stop falling first, not making new lows even with bad news

• Overseas cloud provider earnings, AI commercialization data, and A-share interim reports continuously verify prosperity

• Sustained incremental capital enters during rebounds, rather than single-day pulse-like rallies

Operational advice: Those with heavy positions should first control concentration; don't put all eggs in one basket. Those with light positions shouldn't rush to bottom-fish in one go; wait for signals to improve before entering in batches. Those doing dollar-cost averaging should continue executing plans without letting short-term volatility disrupt the rhythm.

Q4: Is the bull market over?

Conclusion first: The trend of improving prosperity remains unchanged; judging the bull market as over is premature.

To judge if a bull market exists, the core is the earnings cycle. Currently, the overall earnings of A-share listed companies are still on an upward trajectory—in Q1 2026, despite double pressure from a high base last year and exchange rate losses from RMB appreciation, A-shares delivered solid earnings growth, confirming earnings have entered an upward cycle. Institutional estimates show that full-year net profit growth for all A-shares is expected to fall in the 7%-10% range, with non-financial sectors potentially reaching high growth of 20%-25%, where the tech industry is the leading force.

Besides earnings support, three medium-term logics remain unchanged:

① The major direction of economic transformation and upgrading hasn't changed. As the driving role of real estate weakens, the capital market bears multiple roles: direct financing, improving household wealth effects, and boosting consumer confidence. A vibrant stock market is a crucial part of economic transformation, not optional.

② Hard tech breakthroughs rely on capital market support. Semiconductors, AI, commercial aerospace, humanoid robots, solid-state batteries... these directions are in stages with large R&D investments and long return cycles, requiring continuous blood transfusion from the capital market. If the market loses its financing function, not only listed companies are affected, but the entire upstream and downstream industrial chain and employment will feel the pressure.

③ The determination and toolbox for policy-supported markets are ample. On the evening of July 19, China Reform Holdings and China Chengtong, two central SOEs, simultaneously announced share increases, with scales reaching "nearly 10 billion" and "50 billion" levels, respectively. This is another major statement following April 2025. Macro liquidity regulation, long-term capital entering the market, insurance and pension funds increasing equity allocation, and stabilizing forces like Central Huijin are all options in the toolbox. (Data source: Xinhua Net 26.7.20 "Firmly Optimistic About China Reform Holdings and China Chengtong Announcing Continued Share Increases")

Of course,this doesn't mean the market will immediately V-reverse. Oscillating bottom-grinding or even a secondary probe low are possible. But declaring the "bull market over" because of one intermediate adjustment is overly pessimistic. A more rational approach is to retain core positions, control concentration, and wait for earnings, policy, and capital signals to continue verifying.

Q5: Has AI industry prosperity peaked?

In one sentence: Not only has it not peaked, it is accelerating penetration.

Many people equate stock price drops with industry cooling, which is a typical "stock price illusion." Following the transmission chain of "downstream application popularization → midstream model commercialization → upstream computing investment," the three core drivers of the AI industry are currently all moving upward:

Upstream: Computing capital expenditures continue to increase. Capital expenditures for North America's top four cloud service providers are expected to exceed $725 billion in 2026, a YoY growth of over 60%. Meta was previously misread by the market as "selling computing = computing surplus," but soon after announced building a 1GW ultra-large data center in Canada with a total investment of $9.17 billion, directly falsifying the "peak theory." TSMC's Q2 revenue was $40.2 billion, up 36% YoY, with a gross margin of 67.7% hitting a record high, and full-year revenue growth guidance raised to 40%+. If upstream wafer foundry leaders are expanding production, how could demand be poor?

Midstream: Large model commercialization revenue is exploding. Anthropic's ARR (Annual Recurring Revenue) hit $75 billion in July, with monthly growth exceeding $10 billion, expected to achieve single-quarter profitability of $1 billion in Q3, two years ahead of schedule. Domestic players are no less impressive: Zhipu GLM-5.2's ARR broke through $1 billion, growing about 15 times from the beginning of the year, completing its annual target half a year early; KimiK3 saw demand explode within 48 hours of release, forcing a pause in C-end subscriptions due to computing shortages—this indicates from the demand side how strong the willingness to pay for large models is.

Downstream: Application-side Token consumption continues to expand. Although the market begins to worry about the marginal effects of Scaling Law, the overall AI infrastructure is still in an expansion cycle. Cost reductions will instead drive more demand (known in economics as the "Jevons Paradox"); just as faster internet speeds lead to more usage, not less.

What are the real "peak signals" to watch out for? It's when cloud provider capex growth slows significantly, ARR weakens marginally for consecutive months, Token call volumes and unit expenditures see "volume and price double drops," and debt risks of highly leveraged AI companies erupt.None of these signals have appeared yet.

Stock prices can crash due to trading crowding, but industry trends won't reverse just because stock prices fall.

Q6: I bought at highs earlier, now losing money hurts my heart. How can I break even?

First, I want to say: We completely understand the feeling of being trapped after buying at highs. The uncertainty of "when will I break even" is more torturous than the floating loss itself.

Breaking even essentially has two paths; which one you choose depends on your cash flow and holding logic:

Approach 1: Dollar-cost averaging to lower costs, suitable for investors with stable cash flow and long-term optimism.
If you believe the AI industry logic still holds and you have stable income to invest monthly, set a total amount and frequency, invest in batches, and slowly lower your cost basis. When the market rebounds, because your cost is lower, the speed to break even will be much faster than buying once. The essence of DCA is not bottom-fishing, but using time to exchange for space, smoothing out timing anxiety.

Approach 2: Stop-loss and swap positions, suitable for investors with excessive positions, insufficient cash flow, or changed holding logic.
If tech positions are so high they affect sleep, or the reason you bought doesn't hold up (e.g., following rumors), it might be better to actively reduce positions, move funds to directions you are more confident in, and use new gains to cover old losses.

A practical method to judge "stay or go": Compare against the historical maximum drawdown of the funds you hold. If it's a traditional industry fund, a 30%-40% drawdown is basically deep; further downside is limited, so stop-loss isn't very meaningful. But for high-volatility varieties like tech or semiconductors, historical max drawdowns can reach 70%-80%. If you've only drawn down 20%-30% now, seriously assess if you can withstand subsequent volatility.

Core principle: Don't stubbornly hold just because of "unwillingness," nor blindly cut losses just because of "fear of further drops." First figure out how long this money can sit, and whether you believe in this industry, then decide.

Q7: Tech has fallen so much, can I buy the dip and add positions now?

You can consider it, but don't rush in just because "it has fallen a lot."

Before adding positions, ask yourself four questions:

1. Is the industry logic still intact? (Refer to Q5 for answer)

2. Am I buying the right fund direction?

3. Is the proportion of tech in my total portfolio already too high?

4. Is the money for adding positions idle money I won't need for 3-5 years?

If any answer is unclear, don't rush to lower your cost basis.

Based on different risk preferences, here are two executable plans:

Plan A: Pyramid Layout (Suitable for those bullish long-term and able to withstand continued volatility)
If you are bullish on the tech track long-term and can endure continuous oscillation, adopt a gradient batch-addition idea: Use a 20% pullback from the recent rally high as the current baseline, setting two batch-entry intervals.

First Layer Layout Interval: When the index cumulatively draws down to about 30% from its high, representing another 10 percentage points down from current conditions, use 40% of reserved addition funds to buy in batches;

Second Layer Layout Interval: When the index cumulatively draws down to 40% from its high, probing about 20 percentage points lower relative to current prices, use the remaining 60% of 备用 funds to complete additions.

Distributing more funds into deeper correction intervals effectively lowers the overall holding cost, while retaining a basic core position throughout to avoid missing out completely if the market recovers early.

Plan B: Moderate Reduction, Regain Initiative (Suitable for those with excessive positions and floating losses affecting life)
If checking the market daily causes anxiety, better to first lower tech positions to a level where you can sleep (e.g., halve them). Retain half core position + recover half cash: If it continues to fall, the exposure to new losses shrinks; when signals clarify, you still have bullets to catch back; if it rebounds directly, the core position also benefits.

Remember three iron rules:

• Total addition amount must have a cap; don't keep adding as it falls, getting deeper

• When each tier triggers, re-check if the industry logic has changed

• Absolutely do not use loans, emergency money, or short-term turnover funds to add positions

The purpose of adding positions is not to bet on a rebound, but to ensure subsequent volatility always stays within your tolerance.

Q8: I missed out earlier, is it still suitable to enter now?

Whether it's suitable to enter now has nothing to do with whether you missed out earlier; it only depends on current valuations, industry logic, and the investment horizon of this capital.

"FOMO anxiety" is an investment taboo—always thinking "I lost out by not buying earlier, if I don't buy now I'll lose more," then impulsively chasing in, only to get trapped halfway up the mountain.

Objectively speaking, after this adjustment, the valuation pressure on the tech sector has released significantly, and some targets have returned to the "strike zone." But overseas deleveraging, earnings verification, and digesting trading crowding still take time."Fallen a lot" does not equal "the lowest point has arrived."

If you are bullish on tech long-term, currently have low positions, and can lock up capital for three years or more, you can consider starting to build positions, but don't go all in at once. Suggestions:

1. First set a target proportion (e.g., tech accounts for 15% of the portfolio)

2. Invest one-third as an observation position first

3. Deploy remaining funds in batches according to monthly schedules, drawdown magnitudes, or earnings nodes

This way, if the market rebounds immediately, you won't be completely empty-handed; if it continues to fall, you still have ammo to add; you have room to maneuver either way.

In terms of selection, beginners should prioritize broad-based tech indices like the STAR 50, which have wide coverage and relatively dispersed risk; for those who research the semiconductor industry chain and can handle high volatility, consider 细分 directions like chips or semiconductor equipment. The more 细分 the product, the higher the elasticity, but also the higher the holding difficulty.

Finally, ask yourself: Am I buying because of long-term allocation needs, or am I afraid of missing tomorrow's rebound? If it's the latter, halve the amount again.

Q9: Bullish on tech long-term, what proportion of funds is appropriate? How to allocate?

There is no standard answer, but there is a simple and useful "sleep test."

Don't ask "how much for conservative types, how much for aggressive types"; ask yourself a more practical question:If tech falls another 30% in the short term, how much will my entire account lose? Can I still hold normally without rushing to sell?

Here is an intuitive stress test for you:

(Note: Does not include 联动 declines in other equity assets)

Multiply this percentage by the real amount in your account and feel if that number is acceptable. If acceptable, this proportion is reasonable for you; if not, adjust downwards.

Also beware of a pitfall: Looks diversified, actually repetitive. Holding STAR 50, Chip ETF, Semiconductor Equipment, and AI Computing Fund together is not diversification—the underlying holdings may be highly overlapping, all being those tech giants. First calculate the total proportion of tech assets, then set a total cap.The proportion you can hold long-term is the effective position.

Q10: Tech volatility is too high, my heart can't take it. What assets can I pair it with to stabilize?

There is no universal "ballast stone"; different assets solve different problems. First figure out what your portfolio lacks, then match accordingly:

① Worried about betting on a single track → Broad-based index enhancement as foundation
Broad-based indices cover industries more widely, not betting all returns on tech. Historical data shows that after brief market corrections, using STAR Composite Index enhancement etc. to catch rebound opportunities yields good win rates and odds.

② Want to reduce equity volatility and still get dividend income → Dividend/High-dividend assets
Dividend and bank-type assets have low valuations, stable cash flows, and guaranteed dividends, offering certain complementarity to high-valuation growth styles. In a low-interest-rate environment + increased market volatility, operationally stable dividend assets with strong dividend capabilities can serve as the defensive core of the portfolio.

③ Need money short-term, need liquidity → Bonds/Fixed Income +
Bonds and stocks have different return drivers, acting as the account's natural "stabilizer." If there are short-term capital needs, don't let this portion bear tech's high volatility. Fixed Income + products ("bond foundation + small equity enhancement") are also choices for attack and defense, suitable for investors who want to participate in the market but fear big swings.

Recommended combination idea: Broad-based index enhancement foundation + Tech enhancement + Stable asset buffer.
Each asset class has a clear responsibility; rebalance regularly—cash out a bit when it rises too much, add a bit when it falls too much. It's not that more assets are better, but that each performs its duty and dynamic balance is key.

In summary, don't make impulsive decisions when disturbed by emotions.

Reviewing this adjustment, it is essentially aclearing of trading structures, not a reversal of industry logic.

• Fundamentals: AI capex is still accelerating, model commercialization is surging forward, domestic and overseas tech leader earnings continue to beat expectations, and industry prosperity shows no signs of peaking

• Policy Side: Central SOE share increases, long-term capital entering the market, and the strategic positioning of the capital market serving tech innovation support logic remains unchanged

• Funding Side: Stock ETFs saw net inflows exceeding 200 billion in a single week, a record high, with smart money positioning counter-cyclically amidst the crash

The market is a voting machine in the short term and a weighing machine in the long term. Panic emotions can cause stock prices to drop 20% in weeks, but long-term returns are always determined by industry trends and corporate earnings.

For investors, the most important thing now is not guessing where the bottom is, but doing two things well:

First, adjust positions to a level where you can sleep. For heavy/concentrated positions, diversify moderately; for too-low positions, build in batches; for DCA, stick to discipline.

Second, build a "comfortable regardless of rise or fall" portfolio. Keep tech's industry sharpness, use broad-based to capture market beta, use stable assets to buffer volatility, and cooperate with regular rebalancing. This way, when rising you have positions and won't miss out; when falling you have buffers and won't panic.

Investing is a marathon, not a sprint. Every major correction is a test of cognition and mindset.Some cut flesh and leave in panic, others quietly position amidst chaos. Hope you are the latter.

Finally, a phrase for everyone:Bull markets don't rise all the way, bear markets don't fall all the way. Do well in allocation, control emotions, and time will stand on the side of rationality.

Risk Warning: Represents the market views of Guolianan Fund at the time. Markets carry risks; investment requires caution. This material is prepared based on publicly available information deemed reliable by Guolianan Fund. Under no circumstances do the information or opinions expressed in this material constitute investment advice for anyone, nor do they serve as promotional materials or any legal documents. Investors should not regard this viewpoint as the sole reference factor for making investment decisions, nor should they believe it can replace their own judgment. Investors bear all risks and consequences of their investment behaviors themselves. Regular fixed investment is a simple and easy-to-implement investment method that guides investors to conduct long-term investment and average investment costs. However, regular fixed investment cannot evade the inherent risks of fund investment, cannot guarantee investors obtain returns, and is not an equivalent financial management method replacing savings.