GF Securities flags 6 broken A-share rules as AI divergence deepens

Six historical A-share market rules have been broken in 2026, and two more are under threat, as China's equity market undergoes its most structural transformation in two decades, according to GF Securities.

A-shares have posted valuation expansion for three consecutive years for the first time since the exchange opened, GF Securities strategist Liu Chenming wrote in a June 30 mid-year report. The prior ceiling was two years, reached in 2006-07, 2014-15 and 2019-20. The AI-driven bull run in electronics and communication stocks has now entered its fourth year, surpassing the typical two-to-2.5-year lifespan of prior bull market leaders.

"The old map no longer works for the new continent," Liu said. "The rules that held for 20 years were summaries of a China that no longer exists — the property cycle has ended, the demographic dividend has passed its peak, and the profit and market-cap structure of A-shares has fundamentally changed."

The TMT sector's share of total turnover has broken through the 40% ceiling that held from 2019 to 2023, reaching 45% after DeepSeek's launch, Liu noted. The ceiling was 17% during the 2009-10 smartphone cycle and 30% during the 2013-15 mobile internet era. Each technology wave has systematically lifted the threshold.

Institutional ownership of electronics stocks has stayed above 20% for five quarters without triggering a peak — a pattern that previously signaled a top in six historical cases since 2004. The difference this time: the overweight ratio relative to the sector's free-float market cap is actually the lowest on record, because the market capitalization of new-economy stocks has grown so rapidly.

The inventory cycle rule has also failed. When finished-goods inventory growth fell to near zero in June 2023, the historical playbook called for large-scale fiscal stimulus within months. Three years later, no such stimulus has arrived. "This is not policy failure — it is a deliberate shift from 'stimulate at every dip' to strategic restraint," Liu wrote. Cyclical stocks have only rallied on imported inflation from global commodity prices, not on domestic demand.

Two more rules are approaching their breaking point. The concentration of trading in the top 5% of A-share stocks reached 49.8% as of June 24, approaching the 2015 and 2021 highs. Liu pointed to the U.S. experience during the 1990s dot-com cycle, where the top 5% concentration broke through its prior ceiling and has since stabilized near 72%. "When a major technological变革 occurs, the ceiling on concentration itself gets rewritten," he said.

Industry valuation dispersion, measured by the standard deviation of price-to-book percentiles, has hit an all-time high. But Liu cautioned against assuming mean reversion. In six historical cases of extreme dispersion, the peak did not coincide with a bull market top. During the 2020 cycle, elevated dispersion persisted for 20 months, and high-valuation sectors did not underperform when supported by an industrial trend.

The K-shaped divergence between AI and non-AI stocks is unlikely to narrow in the second half. Global cloud capital expenditure continues to be revised upward — the top five providers are expected to spend $769 billion in 2026, up from earlier forecasts. Compute rental prices have risen above $2.80 per hour for Nvidia H100s, while token call volumes at ByteDance's Doubao exceeded 180 trillion daily in June and Google's monthly token consumption approached 3,000 trillion.

On the other side of the K, China's fiscal restraint is sustained partly because external demand provides a buffer. China's major export destinations remain in fiscal expansion cycles, effectively performing demand-pull functions that reduce the urgency for domestic stimulus. Liu said investors should watch two variables for a potential policy shift: the sustainability of fiscal expansion in major economies and the position of the global inventory cycle.

The market is shifting from an incremental to a stock-flow liquidity environment. Net retail inflows reached about 860 billion yuan in the first half, up from 690 billion in the second half of 2025. But the second-half supply-demand gap is expected to narrow by more than 100 billion yuan as margin trading slows, northbound flows face U.S. rate headwinds, and mega-IPOs such as ChangXin Memory Technologies absorb capital.

Beyond AI, Liu identified four sectors with genuine earnings support. Copper and gold have been oversold after a 20%-plus decline following the March U.S.-Iran conflict, with gold's structural thesis intact — global government debt expansion, dollar reserve status erosion, and central bank buying that rose quarter-on-quarter in early 2026. Copper faces a structural supply deficit as mine grades decline and new projects take years to come online, while demand from grid investment, AI data centers and electric vehicles continues to grow.

Energy storage and lithium batteries represent the most confirmed recovery outside AI, with contract liabilities signaling sustained order recovery and June China lithium battery production reaching about 268 GWh — a fourth consecutive monthly record. Innovative drug stocks trade at one standard deviation below their historical mean, with Chinese companies' share of global biotech deals jumping to 70% this year from 50% in 2025. Non-bank financials and fintech IT offer high earnings visibility from sustained capital market turnover.

The one rule that has not broken and will not break, Liu said, is that relative earnings advantage determines relative stock performance. "If the K-shaped fundamentals don't reverse, the K-shaped market won't reverse either," he wrote. "Historical rules are always meant to be broken. But breaking them requires a reason — and that reason is that the era has changed."

This article is for informational purposes only and does not constitute investment advice.