Technological selling point: The first peak has not yet arrived.
```
Since the second quarter, the “New Ning Portfolio” centered on AI technology and the ChiNext Index have risen rapidly, creating extreme divergence with other sectors. The real concern for the market is no longer whether the main theme persists, but rather: with the rise to this level, is the selling point approaching?
Lin Rongxiong, a strategy analyst at Guotou Securities, wrote his core judgment plainly in his strategy theme research published on June 4: “All trends are about group chasing; stocks riding industry trends often end with an M-shaped top. Don’t fret about when the first top appears, mastering the second top is the key.” For the current AI technology, the conclusion is: “The first top still hasn’t arrived.”
The report further distinguishes between “rotation from high to low” and “the first M-top”. Before the industry trend ends, excessive price heat, profit-taking, or policy/fundamental catalysts in low position sectors may all trigger rebalancing from high to low. But as long as the industrial logic remains intact, there is no macro ‘gray rhino’, and the competitive landscape is not broken, the main theme may still return afterward.
AI technology is not without warning signs. The high-to-low index in A-shares is already near its upper boundary, meaning the probability of rotation from high to low is increasing; capital siphoning has become more apparent. But several harder “first top” signals are not yet in place: representative indices have not fallen below the 60-day moving average, leading stocks haven’t seen a final round of clear “valuation surge”, and most companies haven’t overshot their future earnings years above historic top levels.
The first top is often misjudged; the second top is more like the selling point
Industry trend stocks rarely have a clean, sharp top. More common historically is the M-shaped double top: the first is the trading top, the second is the fundamental top.
The trading top often occurs when quarterly earnings growth year-on-year reaches a phase high point. That is, when stock prices first hit a peak, fundamentals usually still look good, earnings are strong, and market sentiment is prone to extrapolating good news. This is why the first top is hard to judge: it's not "bad news led to the drop", but more like "things are so good that they start to loosen after peaking".
The second top is closer to the peak of trailing twelve months (TTM) earnings growth, usually 1–2 quarters apart from the first top, history often shows about half a year. It’s not the perfect escape point, but it’s more identifiable. Except for the bull market in 2015 driven by leverage trading, the second top is usually only 10-15% lower than the first.
Several samples fit this pattern: the 2020–2021 wave of the new energy industry, with the trading top in December 2021, and the second top around June–August 2022; the 2019–2021 consumption upgrade trend, the first top in January–February 2021 and the second in July–August 2021; the technology bull of 2013–2015 and the Nasdaq tech bubble of 2000 show similar M-top structures.
Rotation from high to low is not a top, unless it breaks the trend
Rotation from high to low looks scary: leading stocks adjust, low-position sectors catch up, the market starts questioning the previous industry logic. But when the trend isn’t broken, it’s mostly position rebalancing at the trading level.
To judge if high-to-low rotation has escalated into the first top, the key is to watch a few things.
First, whether it breaks the medium- to long-term moving average. Temporary rotation historically doesn't break the 120-day moving average; if both 60-day or 120-day moving averages are broken, and the 20-day moving average turns downward, the nature changes.
Second, whether macro “gray rhinos” are disturbing the industry trend. In February 2021, the “Mao Index” collapsed, corresponding to tighter regulation and rising U.S. Treasury yields; the “Ning Portfolio” adjusted at the beginning of 2022, corresponding to the Fed’s aggressive rate hike. Simple rotation isn’t a problem; danger comes when macro constraints combine with a broken trend.
Third, whether the low-position sectors really have strong policy or fundamental catalysts. If it’s just profit taking from the highs, the main theme doesn’t necessarily end; if capital keeps migrating and leading stocks start breaking position, then it’s closer to confirming a top.
There are five traces before the first top: best earnings, hottest trading, most expensive leaders
The first top can’t be precisely predicted, but historical peaking phases aren’t entirely without traces.
One common trace is that quarterly earnings growth year-on-year reaches a phase-high. Historical highs of Kweichow Moutai and CATL both show that at the trading top, reported earnings are in their best phase.
The second sign is trading indicators entering extreme congestion. Valuation percentiles, institutional holdings, turnover ratios all heat up together, and the current valuation of leading stocks rises significantly relative to the 10-year average. They don’t tell the exact day to sell, but signal when danger is near.
The third sign is that leaders begin to massively overshoot future earnings. Historically, CATL’s high in 2021 overshot about 3–4 years, while Moutai overshot about 4–5 years; both the “Mao Index” and “Ning Portfolio” near their tops, most constituent stocks overshot by at least three years. Once prices require 3–5 years of high growth to digest, tolerance drops rapidly.
The fourth sign is the rhythm: "large-small-large"—first leaders rise, then it spreads to secondary and tertiary companies or other chain segments, finally leaders lead secondary and tertiary up to the top. CATL before the new energy top, Moutai before the consumption upgrade top, both fit this pattern.
The fifth sign is capital siphoning. When the main line peaks, other indices steadily fall as funds are drawn into the strongest direction. For example, during the “Mao Index” peak, CSI 500 and CSI 1000 weakened; during the “Ning Portfolio” peak, SSE 50 and CSI 300 felt pressure.
For AI: warning signs are appearing, but not enough to confirm the first top
The greatest caution for current AI technology is the rising risk of high-to-low rotation.
The A-share high-to-low index is already close to the upper boundary and is starting to turn. This suggests that the probability of market rebalancing from high-tech to low-valued sectors is rising. Coupled with extreme divergence between AI and other sectors, short-term volatility shouldn’t be underestimated.
But confirmation of the first top requires more evidence. Currently, several key conditions are insufficient.
Representative AI indices’ moving averages are still upward; though some have breached the 20-day moving average, none have breached the 60-day; macro “gray rhinos” have not appeared; other sectors lack clear fundamental catalysts. In other words, current conditions are more like rotation pressure after overheating, not a broken industry trend.
Leaders’ performance isn’t like that of the last sprint pre-top in history. Currently, major leaders haven’t clearly “pulled up valuations”, nor have first-tier leaders outperformed the optical module index; secondary and tertiary companies are actually more active. Historically, the real peaking phase is a final round of valuation surge by the leaders.
Valuation overshooting is also far from historical extremes. Most AI-related companies are not overshooting more than three years, and leading companies’ overshooting is much less than that of the “Ning Portfolio” in 2021 or the first top of the 2015 tech bull.
The only indicator already flashing is the siphoning effect. Funds are focusing on the AI main line, which is a warning signal under the top observation framework, but alone, it’s still insufficient to qualify as the first top.
The real hard constraints are capital expenditure and competition landscape
Technology selling points have two lines.
One is the valuation line. If stock prices overshoot future earnings by more than three years, caution is needed; for the strongest industry trends and top companies, more than five years is near the historical limit. This method is left-sided, possibly missing the last round of valuation surge, but can avoid passive exit from congested positions.
The other is the industry line. Macro “gray rhinos” and industry competition structure are two hard thresholds for AI technology going forward. Global economic recession affects capital expenditure; if the competitive landscape breaks, even if the industry is still booming, it will turn into a selling point.
The framework mentions that in the next half year, high capital expenditure in AI will likely remain unchallenged, but beyond that, future declines are hard to predict; the sustainability of capital expenditure by 2027 will be key. This means that relying solely on industry trends for left-sided sells isn’t easy; most of the time, it’s about playing by ear as circumstances evolve.
As of now, the conclusion is not complicated: short-term high-to-low risk for AI is rising, but evidence for the first top is insufficient. The real turning point is when moving averages break, macro disturbances appear, leading stocks surge valuations, valuation overshoots beyond three years, and fund siphoning all simultaneously become glaring. Historical experience doesn’t guarantee repeat performance, estimates may be off, but this framework at least offers a tougher method than “sell when it’s risen too much”.
Risk disclosure and disclaimerThe market comes with risks. Investment needs to be cautious. This article does not constitute personal investment advice and does not take into account the particular investment goals, financial situation, or needs of individuals. Users should consider whether any opinions, views, or conclusions in this article are suitable for their particular situation. Investing based on this is at your own risk. ```