Money floods AI stocks while Hong Kong has its worst day since March — long Nasdaq, short China tech

Verdict: the AI-bypasses-China divergence looks live, but one entry condition isn't — wait

**Verdict: wait — the divergence is real, but the entry rules aren't all live yet.** The strongest point for the trade is that every KWEB-side condition is already met (price $23.86 versus the $25.80 50-day EMA, RSI 27.3, ADX 24.6) and the news backdrop matches the thesis exactly: per the October 2 Reuters piece, global equity funds drew a second straight week of AI-driven inflows while Bloomberg reported Hong Kong's biggest slump since March. The strongest point against is that QQQ's ADX (14) is 13.9, well short of the required 20, and the five-year backtest won only 41.2% of 17 trades with an 11.0% drawdown — the recent 12-month window (11.2% return, 80% win rate, 1.9% drawdown) may be the outlier, and daily-bar fill approximations make the edge fragile. A near-term mechanical risk: KWEB's $2.096 annual dividend went ex on 2025-12-22 and the next window is typically late December, which can distort the short leg around the ex-date. Also note ownership data on KWEB is thin — one reporter held about 1.9M shares across 5 holders for the June 30, 2026 period, a filed figure that is now dated, not current positioning. The verdict flips to actionable the day QQQ prints a daily ADX (14) at 20 or above with the other conditions intact; it flips to avoid if QQQ closes below $721.07 or KWEB reclaims $25.80 while QQQ's RSI drops below 50. **Conviction breakdown:** thesis support 70, trade readiness 40 (one entry condition short), risk quality 55 (mechanical stops but coarse fills), backtest evidence 55 (long record, weak headline win rate), fundamentals trend 60 (QQQ look-through shows about 29.5% revenue growth and a 53.3% gross margin on its top names; no KWEB issuer snapshot available).

Trade now: divergence is half-live — QQQ's trend strength is the missing trigger

KWEB closed at $23.86, deep below its 50-day EMA at $25.80, with RSI at 27.3 and ADX at 24.6 — so every KWEB-side condition for the divergence entry is already met, and KWEB is sitting right at its nearest support ($24.16) after a 44.3% fall from its range high. The setup is not complete, though: the QQQ leg needs ADX (14) at 20 or above, and it is reading 13.9, about 6.1 points short. QQQ's price ($749.65 vs. its EMA at $721.07) and RSI (71.7, above 50) already pass, so the only missing piece is trend strength on the Q side. The idea argues global AI-driven flows are bypassing Chinese equities; the tape on KWEB supports that half emphatically. Concretely, wait means: hold both legs until a daily QQQ close prints ADX (14) at 20 or above while all other conditions hold — KWEB below $25.80 with ADX above 20 and RSI below 50, and QQQ above $721.07 with RSI above 50. Once triggered, the strategy takes a long QQQ leg and a short KWEB leg with a 2.0% stop-loss on unrealized P&L, a 4.0% take-profit, a 90-bar time stop, and positions capped at 25% each. That stop/target geometry implies roughly 2:1 reward-to-risk per leg. If QQQ instead closes below $721.07, or KWEB reclaims $25.80 while QQQ's RSI drops below 50, the divergence thesis is invalidated — do not chase. Evidence read: the completed backtest on this exact rule set returned 10.8% over five years across 17 trades (41.2% win rate, 11.0% max drawdown), and the trailing 12-month window did better still — 11.2% return over 5 trades with an 80% win rate and just a 1.9% drawdown. One caveat that matters for execution: exits were filled on daily trigger bars, not intrabar, so stop-loss fills are approximate. No robust alternative parameter setup was established, so trade the published rules as written.

Two Flows, One Trade: The Divergence Paid in Every Window Tested

The thesis is a spread trade, not a directional bet: be long the AI-led Nasdaq-100 proxy while shorting Chinese internet via KWEB, capturing the divergence from both sides. The rule set encodes exactly that — it only enters when KWEB trades below its 50-day trend with rising directional strength and momentum under 50, while QQQ sits above its 50-day trend with momentum above 50. That is a double confirmation: China must be actively weak, not just flat, and the AI complex must be actively strong. Per the October 2 Reuters piece, global equity funds have drawn inflows for a second straight week on AI optimism, and per Bloomberg's October 2 coverage, Hong Kong just suffered its biggest slump since March — the news backdrop matches the entry conditions the strategy is built to detect. The completed backtest supports the structure across every window tested. Over 60 months on daily bars, the pair produced a 10.8% total return across 17 trades. The 24-month window returned 7.6% on 7 trades with…

Scores

• Conviction score breakdown: 56
• Thesis support: 70
• Trade readiness: 40
• Risk quality: 55
• Backtest evidence: 55
• Fundamentals trend: 60

Watch items

• QQQ — ADX (14)
• QQQ — Close vs 50-day EMA
• KWEB — Close vs 50-day EMA
• KWEB — RSI (14)
• KWEB — Support level
• QQQ — RSI (14)
• KWEB — Next ex-dividend date