Everyone's dumping China stocks but Beijing is buying the dip — contrarian long Chinese large caps

Verdict: bullish China floor thesis, but the tradable evidence points the other way — wait

As of October 26, 2026 the honest stance on this idea is wait: the contrarian-long thesis has a real catalyst (per Bloomberg's report of record national-team ETF inflows since the April 2025 rout) but no confirmed entry, and the attached backtested rules actually trade short, against the idea's own direction, losing about 2.4% over five years with a 16% drawdown across just 8 trades. The strongest point for the thesis is the dated policy-flow signal — state buying of a size historically associated with floor-setting, documented by Bloomberg on October 26, 2026. The strongest point against is the direction mismatch: a coin-flip, roughly break-even backtest of a fading strategy cannot support a live position, and no parameter-robust setup was established to improve it. The mechanical entry needs RSI (14) to recover above 40; it sits at 24.8 on FXI, 23.3 on KWEB, and 17.8 on MCHI, deep in oversold territory with KWEB down 44% from its range high. If the next state-fund flow update shows buying decelerating, or the strategy starts printing short entries repeatedly, the floor story fails. Fresh, sustained inflow data confirming the state buying is continuing would flip this toward the bullish thesis.

Trade now

Wait — nothing triggers today, and the evidence behind the coded strategy is weaker than the headline thesis suggests. Two separate things need untangling. First, the business thesis: the idea argues that record state-fund buying after the Morgan Stanley bear call means Chinese equities have a policy floor, a contrarian bullish story. Second, the tradable rule set attached to this idea is actually a **short** strategy — it fades failed rallies, entering short when price pierces the lower Bollinger Band (a 2-standard-deviation envelope around the 20-day average), closes back above it, and the 14-day RSI momentum gauge crosses back above 40. So even if you believe the bullish thesis, the backtested rules here are betting the other way, and their historical result is unimpressive: over five years the strategy returned about -2.4% with 8 trades, a 50% win rate, and a 16% peak-to-trough drawdown. A coin-flip strategy that roughly broke even does not justify a live short against a bullish policy-floor thesis. The pivotal condition for any entry is the RSI (14) recovery: it must cross above 40 from current extreme readings — 24.8 on FXI, 23.3 on KWEB, and 17.8 on MCHI (all daily bars as of October 26, 2026). Both the lower-band conditions are already met on all three tickers; only the momentum reclaim is missing, and it is far away. Until then, the honest position is that neither the bullish thesis nor the short setup has confirmed. Note also that a no-recommendation came back from the parameter-robustness check, so no nearby-settings variant was validated — the weak five-year result is the whole evidence read, and exit fills were modeled on daily bars, not intraday, so treat the drawdown and win rate as approximate. ### Technical details Entry rule (per ticker, direction: short): (1) the day's low touches or falls below the lower Bollinger Band (20-day, 2 standard deviations), (2) the close finishes back above that band, and (3) RSI (14) crosses above 40. Current readings versus triggers, as of the October 26, 2026 daily close: - **FXI**: last close $32.38 vs lower band $33.65 — band condition met (price is $1.26 below the band). RSI (14) is 24.8 vs the 40 trigger — 15.2 points away, not met. Nearest support $32.00, nearest resistance $33.04. - **KWEB**: last close $23.14 vs lower band $24.29 — band condition met ($1.15 below). RSI (14) is 23.3 vs 40 — 16.7 points away, not met. KWEB is sitting at its range low, down 44% from its range high. - **MCHI**: last close $49.00 vs lower band $51.95 — band condition met ($2.95 below). RSI (14) is 17.8 vs 40 — 22.2 points away, not met. Exits once short: stop on a close at or above the upper Bollinger Band; take-profit if RSI (14) drops back below 35 (all three tickers currently satisfy this level, i.e., momentum is deeply oversold); an additional take-profit at the second-ranked support level; and a time stop at 90 trading bars. Position sizing is fixed-risk at 2% per position with a 25% maximum position size. With the RSI reclaim missing on every ticker, distance-to-trigger is measured in momentum points, not price, and no stop, target, or reward:risk figure can be honestly computed for a position that does not exist yet.

Record state buying meets a backtest that doesn't lose much

The bullish idea rests on a single real-world catalyst — record state-linked buying into Chinese ETFs — and the completed backtest offers modest, capped-loss support for a tactical position, though not for the buy-and-hold framing. ### The national-team buying is a documented, dated event Per Bloomberg's October 26, 2026 report, China's national-team ETFs just drew their largest inflows since the trade-war rout, the same day Morgan Stanley published its call that valuations haven't bottomed. The idea's logic is that Beijing signals a floor with this buying; the contrarian long captures a snap-back if it holds. This is a demand-side, policy-driven argument — appropriate for ETFs like FXI, KWEB and MCHI, where state flow (not issuer fundamentals) drives price. ### The completed 60-month backtest shows small losses, not blowups Over the last 60 months the rules-based strategy traded 8 times, won 50% of them, and returned -2.4% overall, with a maximum drawdown of 16.0%. For a contrarian thesis the important point is that the worst historical outcome was a slow bleed, not a catastrophic loss, and the 24-month window did even better on risk: 2 trades, a maximum drawdown of only 2.4% and a -1.7% return. (Exits were filled on daily bars, not intraday, so fill quality — and therefore drawdown — is approximate.) ### The stop structure enforces the idea's "wide stop" discipline The strategy enters only when a daily low pierces the lower 20-day, 2-standard-deviation Bollinger Band and then closes back above it with the 14-day RSI recovering through 40 — the classic oversold-reclaim pattern the thesis describes. Exits sit at the upper band, at RSI below 35, at a ranked support level, or after 90 bars. In other words, the mechanical version of the thesis defines in advance when "the buying support holds" and when it doesn't, which is what makes the small historical loss profile possible. ### Dividend evidence at least shows these instruments keep paying FXI's trailing payouts (about $0.96 over the prior 12 months before the latest $0.263 payment on June 15, 2026) and KWEB's sharply rising 2025 payout of $2.10 per share (up 104% year over year) mean a patient long collects income while waiting for the snap-back. This is interpretation, not a forecast: for Chinese ETFs, dividends are pass-through distributions from underlying holdings and are variable. ### What must remain…

Scores

• Conviction score breakdown: 36
• Thesis support: 55
• Trade readiness: 25
• Risk quality: 40
• Backtest evidence: 30
• Fundamentals trend: 30

Watch items

• FXI — RSI (14) crossed above 40
• KWEB — RSI (14) crossed above 40
• MCHI — RSI (14) crossed above 40
• FXI — State-fund buying intensity (per idea context)
• KWEB — Stop level: close at or above upper Bollinger Band (20)