To The Moon

A review of historical data for the ChiNext Index and STAR 50 Index reveals that in extreme cases where monthly declines exceeded 15%, the probability of a recovery in the following month or two months is relatively high, with rebound strength often surpassing market expectations. With both indices' July declines already entering historical extreme territory, historical patterns suggest that the period from August to September could be a relatively high-probability window for recovery.

Historical Extreme Decline Review

The three largest monthly declines in the history of the ChiNext Index are: January 2016, down 26.53%; July 2026, down 23%; and August 2015, down 21.38%. The three largest monthly declines for the STAR 50 Index are: July 2026, down 25.9%; January 2024, down 19.62%; and March 2026, down 15.57%. Note: Data sourced from Wind. The ChiNext Index was launched on June 1, 2010, with a base date of May 31, 2010. Its annual returns from 2021 to 2025 were +12.02%, -29.37%, -19.41%, +13.23%, and +49.57%, with annualized volatility of 26.62%, 26.92%, 18.06%, 37.22%, and 25.32% during the same period. The STAR 50 Index was launched on July 23, 2020, with a base date of December 31, 2019. Its annual returns from 2021 to 2025 were +0.37%, -31.35%, -11.24%, +16.07%, and +35.92%, with annualized volatility of 21.46%, 26.50%, 19.83%, 36.36%, and 27.34%. Index constituent stocks are adjusted periodically according to index compilation rules, and backtested historical performance does not predict future index movements.

Similarities and Differences Between This Decline and Past Ones

Similarities: All major declines occurred after significant prior rallies, with valuations at elevated levels and accompanied by external shocks—such as the 2015 deleveraging, the 2016 circuit breaker mechanism, the 2024 macroeconomic liquidity pressure, and the 2025 geopolitical conflicts. This current decline is similarly built on a rapid rise in AI-related sectors from April to June, with price-to-earnings ratios hitting extreme levels, triggered by Meta's sell-off of computing power, which sparked concerns over overcapacity, along with the transmission of the Philadelphia Semiconductor Index (SOX) plunge. Differences: A key distinction is that the 2015/2016 events were systemic liquidity crises caused by institutional shocks and deleveraging, while the 2024/2025 downturns were driven by macroeconomic conditions and the semiconductor cycle decline. This current decline is more about short-term supply-demand expectation fluctuations in the industry plus valuation bubble digestion, without reversing the long-term AI trend (cloud capital expenditures, computing power demand). If subsequent earnings reports confirm robust computing power demand, the recovery logic for this event may be clearer than in most historical cases.

Pattern Summary

Pattern One: High probability of a rise in the following month. In four data sets, the probability of an increase in the next month is 75% (only the 2016 circuit breaker led to continued bottoming), and both STAR 50 instances saw gains of over 17% in the following month. Even if the next month continues to decline (January 2016), the third month sees a strong rebound. Pattern Two: The greatest elasticity occurs in the third month. After two extreme declines in the ChiNext Index, the average gain in the T+2 month was as high as +18%, indicating that after extreme sentiment clears, medium-term recovery momentum is very strong. Pattern Three: The trend after a major decline unfolds in two stages. Stage one (within one month): There may be continued inertia bottoming or a moderate recovery. Stage two (within two months): A significant rebound is highly likely. Pattern Four: The nature of the decline determines the recovery slope. Liquidity/sentiment-driven declines (August 2015, January 2024): Rapid recovery in the next month, with high elasticity. Institutional/structure shocks (January 2016 circuit breaker): Continued bottoming in the next month, with a strong recovery only in the third month. An intact industry logic plus valuation digestion (this time): Falls between the two, with a relatively high probability of a recovery window in August-September.

Implications for the Current Market Situation

In July 2026, the ChiNext Index fell by approximately 23% (the second largest in history), and the STAR 50 Index fell by approximately 25.9% (the largest in history), both entering historical extreme territory. Based on historical patterns, August-September has a relatively high probability of being a recovery window. This decline is more about short-term supply-demand expectation fluctuations in the AI industry plus valuation digestion, without reversing the long-term industry trend (cloud capital expenditures, computing power demand). This is fundamentally different from the 2015 deleveraging and 2016 circuit breaker institutional shocks, and the recovery logic appears clearer than in most historical cases.

Key Subsequent Verification Points

Regarding key subsequent timing points, the following are outlined: First, the four major cloud service providers—Google, Microsoft, Meta, and Amazon—will sequentially release their quarterly reports, with a focus on the impact of capital expenditures on fundamentals. Second, on August 7, the US July non-farm payroll and unemployment rate data will be released, which will influence the pace of rate cuts and liquidity expectations. Third, on August 26, Nvidia's earnings report will be disclosed, with attention needed on its Beta acceleration. In terms of allocation direction, the AI computing power industry chain, represented by optical modules, remains one of the core themes of the ChiNext Index. After the deep adjustment in July, the valuation pressure on the sector has been digested to a considerable extent. Based on historical patterns and the pace of sentiment clearing, August to September could theoretically be the most significant window for recovery. For medium-to-long-term positioning, phased attention to allocation opportunities may be advisable. In terms of investment tools, the ChiNext AI ETF (159363) and its OTC feeder funds (Class A 023407, Class C 023408) focus on optical module CPO leaders, with the target index containing approximately 40% of Zhongji Innolight, Eoptolink Technology, and TFC Communication, serving as a core AI computing power flagship. Additionally, the ChiNext AI ETF (159363) has a scale exceeding 6.3 billion yuan and an average daily trading volume over the past six months of over 1 billion yuan, leading in scale and liquidity among the eight ETFs tracking the same index. Data sources: SSE, SZSE, Wind, etc. Reminder: Recent market fluctuations may be significant, and short-term gains and losses do not predict future performance. Investors must prudently invest based on their own capital situation and risk tolerance, paying close attention to position and risk management. Fee Note for ETF: When investors subscribe or redeem fund shares, the subscription and redemption agency may charge a commission of up to 0.5% of the standard. On-exchange trading fees are subject to the actual charges of the securities company, with no sales service fee charged. Fee Note for Feeder Funds: The ChiNext AI ETF Feeder Fund Class C charges no subscription fee; the redemption fee is 1.5% for holdings within 7 days and 0% for holdings of 7 days or more; the sales service fee is 0.3%. The ChiNext AI ETF Feeder Fund Class A subscription fee is 1% for amounts under 1 million yuan, 0.6% for amounts between 1 million yuan (inclusive) and 2 million yuan, and 1,000 yuan per transaction for amounts of 2 million yuan (inclusive) or more; the redemption fee is 1.5% for holdings within 7 days and 0% for holdings of 7 days or more; no sales service fee is charged. Risk Warning: The ChiNext AI ETF passively tracks the ChiNext AI Index, with a base date of December 28, 2018, and a launch date of July 11, 2024. Index constituent stocks are adjusted periodically according to index compilation rules, and backtested historical performance does not predict future index performance. The index constituent stocks mentioned in this article are for illustration only. Individual stock descriptions do not constitute any form of investment advice and do not represent the holdings or trading activities of any fund under the management. According to the fund manager's assessment, the risk level of the ChiNext AI ETF is R4 (Medium-High Risk), suitable for aggressive (C4) and above investors. Please refer to the sales institution for suitability matching opinions. Any information appearing in this article (including but not limited to individual stocks, comments, forecasts, charts, indicators, theories, and any form of expression) is for reference only. Investors must be responsible for their own investment decisions. Furthermore, any views, analyses, and predictions in this article do not constitute investment advice to readers and are not responsible for any direct or indirect losses arising from the use of this content. Fund investment carries risks. Past performance of a fund does not represent its future performance. The performance of other funds managed by the fund manager does not constitute a guarantee of the fund's performance. Fund investment should be approached with caution. MACD golden cross signal forms, these stocks are rising well!