〔Overnight Brief〕STAR 50 Jumps 6.39% on Growth Rotation, but Shenwan Maps 2026 Remainder to Consolidation as Fluid-Dynamic Test Rejects the Momentum Equation — Mid-Cap Beta vs Missing Structural Anchor

0. Weekly Arc

The A-share week closed with growth-style leadership: STAR 50 jumped 6.39% and ChiNext rose 1.52%, while SSE 50 dipped 0.26% and CSI 300 edged down 0.06% ^[1]. The mid/small-cap complex outperformed—CSI 500 +2.89%, CSI 1000 +3.51%, CSI 2000 +3.86% ^[1]. Yet Shenwan Hongyuan strategy tags the rebound as small-scale and maps the remainder of 2026 to a "consolidation/resting period," on the view that neither tech nor non-tech can rebuild a major-wave structural mainline while domestic growth, the policy mix, and overseas asset allocation all lack upward elasticity ^[2]. The asymmetry: a tactical growth rotation against a missing structural anchor ^[2][1].

1. Strategy and Policy Frame

• **[ESCALATED] Mid-term unchanged — Shenwan Hongyuan strategy:** writes that the September Fed rate hike plus further hawkishness has not derailed a short-term AI-chain rebound, but the mid-term outlook is unchanged and the stage without a breakthrough direction persists ^[2].
• **[ESCALATED] Tech preconditions unmet — Shenwan Hongyuan strategy:** tech needs to wait for heavyweight industry catalysts to rebuild consensus—ARR growth space reopening, capex expectations rising, and computing-inflation expectations returning; GPT-6 fits the post-2025 catalyst pattern but is already proven insufficient to drive tech fully stronger ^[2].
• **[ONGOING] Overseas allocation pressure — Shenwan Hongyuan strategy:** frames high US Treasury yields as a mid-term problem—weak US fiscal discipline, central-bank reserve-asset diversion, and Japan economic-cycle fragility—without a matching mid-term policy response; argues Fed hikes are not a solution and may worsen mid-term risk concerns ^[2].
• **[NEW] Domestic setup — Shenwan Hongyuan strategy:** domestic Q3 2026 is marginally better than Q2 2026, but A-share earnings validation may be deferred to 26... (truncated in source) ^[2].

2. Market Data and Style Read

• **[NEW] Broad-based indices, week ended Sep 19:** SSE Composite +0.61%, Dongcai All A +1.05%, CSI 300 -0.06%, ChiNext +1.52%, STAR 50 +6.39%, SSE 50 -0.26% ^[1]. Mid/small caps (CSI 500/1000/2000) outperformed large caps ^[1].
• **[NEW] Industry leaders/laggards:** electronics, communications, real estate led; petrochemicals, agriculture/forestry/fishery, coal lagged; 16 of 31 SW industries posted gains ^[1].
• **[NEW] Earnings-revision leaders:** coal, retail, real estate, defense/military, banking ^[1].
• **[NEW] Industry prosperity:** industrial metals prices broadly rose, energy/chemical prices adjusted to varying degrees, lithium-battery upstream materials notably weakened, chip prices rose ^[1].
• **[NEW] Valuation extremes:** electronics, communications, machinery equipment elevated; beauty care, food/beverage, non-bank finance subdued ^[1].
• **[NEW] Margin-financing inflows (top):** electronics, power equipment, building materials ^[1].
• **[NEW] Trading heat:** TMT heat notably rose, major-consumer cooled ^[1].
• **[NEW] Allocation cues:** communications, electronics, computing on growth-end prosperity and capital return; industrial metals, wind-power equipment, chemical raw materials, and marine equipment around earnings-revision and prosperity-up clues ^[1].

3. Capital-Flow Fluid Dynamics

• **[NEW] Framework, design — Shenwan Hongyuan research:** translates the Navier–Stokes equations into an A-share capital-flow framework—industry share as density, capital migration as velocity field, free-float market cap as effective cross-section, valuation overhang as pressure, trading friction as dynamic viscosity ^[3]. Conservation law: net inflow takes the −4(cross-section) identity form; the momentum equation yields upwind-convection and viscous terms; the pressure Poisson equation delivers implied pressure imbalance (F9); the inviscid-limit Kelvin circulation theorem delivers the strongest test, b=1 ^[3].
• **[NEW] Sample and capacity calibers:** SW secondary industry daily panel, 2016/12 to 2026/9, 125 industries and 259,695 industry-days; shock constructed from net active buys; capacity tested across free-float market cap, free-float share count, and constituent-stock count, with all estimators calibrated on synthetic data before being read on the real panel ^[3].
• **[NEW] Kinematic part holds:** net inflow scaled by capacity (g=F/A) at h=20 posts IC 0.026 (t=4.1); unscaled 0.012 (t=2.4); scaled by turnover 0.014 — meaning the asset base, not amount size or trading activity, is what carries the signal ^[3].
• **[NEW] Dynamic part fails:** the headline finding is that the kinematic part of the fluid analogy stands while the dynamic/momentum part does not, capping the framework's predictive reach ^[3]. Pressure-gradient factor F3 IC reads 0.0... (truncated in source) ^[3].

4. What Would Falsify It

• The bull case requires Shenwan's prerequisites—a heavyweight AI beta catalyst, ARR growth space reopening, capex expectations rising, and computing-inflation expectations returning—and none is currently visible; GPT-6 is described as already tested and insufficient ^[2].
• The bear case runs through the late-September capital-market expectations window, where tech's volatility-adjustment leg may extend ^[2]. Earnings-validation timing is the second open thread, with the broker's Q3-vs-Q2 framing conditional on a 26... print that is truncated in the source ^[2].
• Source-quality flag: the strategy view rests on a single weekly note and the capital-flow result on a single research piece; the momentum-equation rejection in ^[3] is the headline, but the F3 IC and the Q3 earnings-validation date are both truncated in the source material and should be re-checked before any model deployment ^[2][3].