Since the second quarter, the "New Ning Portfolio" centered on AI technology and the ChiNext Index have surged rapidly, widening the divergence with other sectors to an extreme level. The market's true concern is no longer whether the main trend exists, but rather: having risen to this point, is the selling point approaching?

Guotou Securities strategist Lin Rongxiong stated in a thematic research report released on June 4th, presenting the core judgment directly: "All trends are essentially group behavior." Stocks driven by industrial trends often conclude by forming an M-shaped double top. "Don't agonize over when the first peak appears; the key to success lies in identifying the second peak." Regarding the current AI technology sector, the conclusion is that "the first peak has yet to arrive."

The report further distinguishes between "rotation from high to low valuation" and "the first peak of an M-top." Before an industrial trend ends, factors like temporary overheating in stock prices, profit-taking, or policy/fundamental catalysts in lower-valuation sectors can trigger a rebalancing from high to low. However, as long as the underlying industrial logic remains intact, without major macroeconomic shocks or a collapse in industry competitive dynamics, the main trend is likely to return.

Current AI technology is not without warning signs. The A-share high-to-low rotation index is nearing its upper bound, indicating an increased probability of such rotation. Capital concentration effects are also becoming more pronounced. However, several stronger signals for a "first peak" are not yet fully aligned: representative indices have not broken below their 60-day moving averages; major leaders have not shown a final significant round of "valuation expansion"; and most companies have not exceeded the level of future earnings透支 seen during historical peak phases.

The First Peak is Often Unclear, the Second Peak is More Like a Selling Point

The peak for industrial trend stocks is rarely a sharp, clean top. Historically, an M-shaped double top is more common: the first is a trading peak, the second is a fundamentals peak.

The trading peak often occurs when quarterly year-on-year profit growth is near its cyclical high. In other words, during the first price surge to a peak, fundamentals typically still appear strong, earnings remain impressive, and market sentiment is most prone to extrapolating good news linearly. This is why the first peak is hard to judge: it's not a case of "bad news arrives, so it falls," but more like "it starts to weaken after reaching a point that seems as good as it can get."

The second peak aligns more closely with the high point of TTM profit growth, usually occurring 1-2 quarters after the first peak, historically often around six months apart. It's not the perfect exit point at the absolute top, but it's more identifiable. Excluding the 2015 leverage-driven bull market, the second peak is typically only 10%-15% lower than the first.

Several examples follow this pattern: the new energy industry wave of 2020-2021 saw a trading peak in December 2021, followed by a second peak around June-August 2022. The consumption upgrade theme of 2019-2021 had its first peak in January-February 2021 and its second in July-August 2021. The 2013-2015 technology bull market and the 2000 Nasdaq dot-com bubble also exhibited similar M-top structures.

Rotation is Not a Peak Unless it Turns into a Trend Breakdown

High-to-low rotation can look alarming: the leading high-valuation sector adjusts, lower-valuation sectors rally, and the market begins to question the original industrial logic. However, when the industrial trend is not broken, this is more about portfolio rebalancing at the trading level.

Judging whether rotation has escalated into the first peak depends on several key factors.

First, whether key moving averages are broken. Pullbacks caused by temporary rotation historically often don't break the 120-day moving average. If the 60-day or 120-day moving average is broken, accompanied by the 20-day moving average turning downward, the nature of the move changes.

Second, whether major macroeconomic shocks begin to disrupt the industrial trend. The collapse of the "Mao Index" in February 2021 corresponded with tightening regulation and rising U.S. Treasury yields. The adjustment of the "Ning Portfolio" in early 2022 corresponded with the Fed's aggressive interest rate hikes. Pure rotation is not the issue; the danger lies when macroeconomic constraints combine with a breakdown of the trend.

Third, whether lower-valuation sectors have genuinely strong enough policy or fundamental catalysts. If it's merely profit-taking in high-valuation stocks, the main trend doesn't necessarily end. If capital migration persists and leading stocks in the main trend begin to break down, it is closer to confirming a first peak.

Five Signs Before the First Peak: Peak Earnings, Peak Trading Heat, Peak Valuations for Leaders

The first peak cannot be precisely predicted, but historical run-up phases are not completely without clues.

One common sign is quarterly year-on-year profit growth reaching near its cyclical high. Historical highs for Kweichow Moutai and Contemporary Amperex Technology Co. Limited (CATL) show that when the trading peak appears, reported earnings are often still at their most impressive stage.

A second sign is trading indicators reaching extreme crowdedness. Valuation percentiles, institutional holdings, trading volume share, and other indicators simultaneously heat up, with core leaders' current valuations significantly elevated relative to their 10-year historical average. They can't pinpoint the exact day to sell but can signal the danger zone is approaching.

A third sign is core leaders beginning to significantly price in future earnings. Historically, CATL's 2021 high priced in about 3-4 years of future earnings, while Kweichow Moutai priced in about 4-5 years. Near their peaks, most components of the "Mao Index" and "Ning Portfolio" priced in over three years of future earnings. Once a stock price requires 3-5 years of high future growth to justify, the margin for error rapidly declines.

A fourth sign is the market rhythm following a "big-small-big" pattern: first, core leaders rise, then the rally spreads to second/third-tier companies or other segments of the industrial chain, and finally, large leaders lead the second/third tier in a final charge to the peak. This pattern was seen with CATL before the new energy peak and Kweichow Moutai before the consumption upgrade peak.

A fifth sign is capital concentration. When the main trend charges towards a peak, other indices decline continuously as capital is sucked into the strongest direction. For example, during the "Mao Index" peak phase, the CSI 500 and CSI 1000 weakened. During the "Ning Portfolio" peak phase, the SSE 50 and CSI 300 came under pressure.

Comparing to AI: Warning Signs Have Appeared, But Not Yet Confirming a First Peak

For current AI technology, the most critical warning is the rising pressure for high-to-low rotation.

The A-share high-to-low rotation index is nearing its upper bound and showing signs of inflection. This indicates an increasing probability of market rebalancing from high-valuation tech to lower-valuation sectors. Coupled with the extreme divergence between the AI main trend and other sectors, short-term volatility should not be underestimated.

However, confirming a first peak requires more evidence. Several key conditions are not yet fully met.

Representative AI indices still have upward-trending moving averages; although there have been breaks below the 20-day average, the 60-day average has not been broken. Major macroeconomic shocks have not yet appeared. Other sectors also lack particularly clear and strong fundamental catalysts. In other words, the current situation resembles rotation pressure after an overheated main trend, rather than a broken industrial trend.

The performance of leading stocks also does not resemble the final sprint before a historical first peak. Currently, major leaders have not shown significant "valuation expansion"; even first-tier leaders have not outperformed the optical module index, while second/third-tier companies are more active. According to historical experience, the true final charge phase usually involves large leaders pulling up valuations one last time.

The degree of valuation based on future earnings has also not reached the extreme levels seen at historical peaks. Currently, most AI-related companies price in no more than three years of future earnings. The透支 level for core enterprises is far lower than during the first peak of the 2021 "Ning Portfolio" or the 2015 technology bull market.

The only signal already flashing is the capital concentration effect. Capital flowing into the AI main trend is relatively obvious, which is a warning signal within the peak observation framework. However, this single factor alone is insufficient to classify the current situation as the first peak.

The Real Hard Constraints are Capital Expenditure and Competitive Dynamics

There are two lines of thinking for selling technology stocks.

One is the valuation line. When a stock price prices in over three years of future earnings, caution is warranted; for the strongest industrial trends and best companies, pricing in over five years has historically been near the limit. This selling approach is more left-sided; it might miss the final "valuation expansion" phase but can avoid being forced to exit during crowded conditions.

The other is the industrial line. Major macroeconomic shocks and industry competitive dynamics are two hard thresholds for AI technology going forward. Whether the global economy enters a recession will affect capital expenditure. If competitive dynamics collapse, even if the industry remains highly prosperous, it could become a selling point.

The framework notes that high AI capital expenditure is difficult to disprove over the next half year, but the timing of a more distant slowdown is hard to predict accurately. The sustainability of capital expenditure in 2027 will be a key observation window. This also implies that relying solely on the industrial trend for left-sided selling is not easy; more often, one can only assess the situation as it develops.

Applying this to the present, the conclusion is not complicated: short-term high-to-low rotation risk for AI technology is rising, but evidence for the first peak is still insufficient. The moment that truly warrants a change in judgment would be when signals like moving average breakdowns, macroeconomic disturbances, valuation expansion in leaders, valuation透支 exceeding three years, and capital concentration all become glaringly apparent simultaneously. Historical experience doesn't guarantee future repetition, and calculations can have errors, but this framework at least provides a more robust judgment method than simply "selling because it has risen too much."