Hong Kong — End-of-Day Verdict
2 ACQUIRE (both QUEUED) + 2 DISPOSE + 0 ANTICIPATE
Recommendations
2 to acquire · 2 to dispose
| Ticker | Name | Close | Action | Action Price Where to act — a buy-limit on a pullback or a buy-stop above resistance, not the current price. Sanity-check vs Close: on a held name, an action price stranded above the close means the queued level was set when the stock was higher — stale, awaiting reprice. | Stop The risk-defined exit, set below the entry for a long (~1.2–1.8×ATR). A held long whose Stop sits at/above the Close is already below its stop — a stale box awaiting reprice. | 1wk Model estimate of the likely price at this horizon — a forward projection, not a guarantee, and separate from the trade's Action Price or Stop. Where the ML estimator is enabled these are per-ticker calibrated forecasts (~80% coverage band) that are refreshed periodically and can drift between refreshes; otherwise they're interpolated from the signal's swing target. Confirm against the settled close. | 4wk Model estimate of the likely price at this horizon — a forward projection, not a guarantee, and separate from the trade's Action Price or Stop. Where the ML estimator is enabled these are per-ticker calibrated forecasts (~80% coverage band) that are refreshed periodically and can drift between refreshes; otherwise they're interpolated from the signal's swing target. Confirm against the settled close. | 12wk Model estimate of the likely price at this horizon — a forward projection, not a guarantee, and separate from the trade's Action Price or Stop. Where the ML estimator is enabled these are per-ticker calibrated forecasts (~80% coverage band) that are refreshed periodically and can drift between refreshes; otherwise they're interpolated from the signal's swing target. Confirm against the settled close. | Rationale |
|---|---|---|---|---|---|---|---|---|---|
| 0857 | PETROCHINA CO LTD H SHS | 10.19 | BUY — QUEUED @10.19 | 10.19 | 9.85 | 10.17 (-0.2%) | 10.19 (+0.0%) | 10.23 (+0.4%) | PetroChina is the cheapest quality name in the book right now: it screens UNDERVALUED with a 20% margin of safety, it is the only new-entry candidate that clears the fundamental quality screen, and the shares still trade below independent estimates of fair value after their recent run. The technical case caught up this session - three consecutive STRONG BUY prints took it to the top of the buy ranking, and it is trading right at the model's entry level rather than extended above it, so we are not chasing. We are buying with a limit rather than at market because interim results land on the same session, and the sensible way to own an earnings reaction is to let the price come to you. Ops: (board lot 2,000) = HK; sizing 0.7155 x 600k x 20% x 1.25 UNDERVALUED modifier. L6 no gap-chase and horizon stance queue-below are why this is gtc_limit at the printed Entry_Price 10.19, never market. HK-1 chase-family conditions verified met: RSI 61.0 < 70, 3 consecutive prints, Entry_Distance_ATR 0.00. Adj conviction 86 +4 val -2 bias = 88. Forward Est_* are flat (+0.4% at 12wk) - the case is signal + valuation + quality gate, not the projection. Expires 2026-09-02. · news |
| 0939 | CCB - H SHS | 9.14 | ADD — QUEUED @9.03 | 9.03 | 8.68 | 9.09 (+0.7%) | 9.21 (+2.0%) | 9.25 (+2.4%) | China Construction Bank has pulled back into the level where this book has historically bought it best - a dip toward its volume-weighted average price while the trend underneath stays intact - and conviction has been climbing steadily all session rather than spiking. It is the most reliable buy pattern in this portfolio's track record. We are placing a limit below the current price rather than buying at market: interim results and the new Fed Chair's first Jackson Hole keynote both land before this can fill, so the order is deliberately structured to execute only if the shares come to us on weakness, with a defined stop just below. If the reaction is positive, the order simply does not fill and costs nothing. Ops: (board lot 1,000) @ 9.03 = HKtakes. Entry_Distance_ATR -0.62 = close is past entry, so no-chase mandates queue-not-market. PullbackBuy_VWAP live bias +2 (n=13) per HK-7 refresh; L1/HK-2 preferred cell. NA_CCY so val_adj 0 and quality gate off - printed MoS +0.21 NOT credited across the CNY mismatch. Horizon stance on 0939 is tighten-trail (advisory against adding) - accepted only because a limit-below can only fill into weakness and risk is bounded at HKby the 8.68 stop. Expires 2026-09-02. |
| 0522 | ASMPT | 160.90 | SELL 85.46% | 160.90 | 155.02 | 165.13 (+2.6%) | 171.48 (+6.6%) | 168.92 (+5.0%) | ASMPT has been rallying inside a downtrend and the model flipped it to a strong sell at the closing bell - the single most reliable exit pattern this portfolio has recorded. On top of that the shares screen as extremely expensive against any reasonable estimate of what the business is worth, and they fail the fundamental quality screen. We are down 18% on the position and choosing to take the loss on the bulk of it rather than defend it, while deliberately keeping a small stub so we still participate if the semiconductor cycle data due next week comes in strong. Ops: of 1,100 (board lot 100) = HKproceeds, realizes -HKvs 196.35 cost; retains 200. Late flip (AVOID x4 -> STRONG SELL on the settled print) - L4/HK-13 settled close is authoritative; HK-9 says trust a HK exit at the point of signal (1-week instrument), live bias +3 (n=19, hit_1w.632). VALUATION AMPLIFY DECLINED: MoS -4.33 <= -3 would bump 85% -> full exit, refused on HK-9 (no full exit justified on the 4wk view) plus the 09-01 Korea semi-exports hold-through stance. Valuation amplified rank, not size. Est_* all sit ABOVE the exit price (+2.6% at 1wk) and argue against - down-weighted because calib_1w is -0.261. Stop on residual HELD at 155.02, NOT lowered to the model's 144.11. Not an HK-11 re-issue: 08-25/08-26 were plain HOLDs. |
| 3750 | CATL | 616.50 | SELL 30% | 616.50 | 600.00 | 615.40 (-0.2%) | 621.75 (+0.9%) | 632.51 (+2.6%) | CATL has gone nowhere for us. The model has refused to take a view on it for six straight readings - its internal indicators are in outright conflict - while the position sits 18% underwater and the twelve-week outlook has flattened to roughly today's price. A holding that the system cannot form an opinion on, at a loss, with no projected upside, is dead capital. We are cutting a third of it at the opening bell and putting the rest on a hard line, rather than continuing to carry a decision we keep deferring. Ops: of 300 (board lot 100) HKrealizes -HKvs 751.80 cost; retains 200. TWO independent triggers: HK-11 escalation (3rd consecutive unexecuted re-proposal - 08-19 withdrawn, 08-25, 08-26 - rule mandates ONE market-at-open escalation then permanent retirement) AND ADR-0012 armed + tick (d) 6 consecutive NO TRADE -> profit_take_1 30%. Tick (c) narrowly missed (632.51 vs 628.83 threshold). NoTrade_Reason verbatim: 'Consensus conflict + low conviction'. NoTrade|sell bias 0 (n=14) - no credit, this is execution discipline not conviction. Market order accepts the 08-31 BYD read-through (+/-7% two-sided, ~+/-HKon) as the cost of honouring HK-11; that same event is why the residual 200 is retired rather than exited. RETIRED after this fill - re-arm ONLY on close < 581.50 (close - 2.0 x ATR). Standing stop 600.00 held. |
| 0005 | HSBC HDGS(USD0.50) HK O'SEAS | 161.40 | HOLD | — | 159.50 | n/a | n/a | n/a | We are holding HSBC and leaving the protective stop where it is rather than taking profits. The model's profit-taking trigger did fire, but it fired on a technicality that is misbehaving across this entire portfolio this week, and the new Fed Chair delivers his first major policy speech before we could act - an event that moves bank shares more than anything else on the calendar. Our downside is already tightly controlled: the stop sits just over 1% below the current price. We would rather let the event happen and then decide. Ops: FIRED armed + 2 ticks (b) cv = -24, (c) T12W 160.31 < 164.63 -> would be a 50% trim = HKDEFERRED on four grounds: (i) PROFIT_ARMED is the T12-progress artifact and Avg_Cost is BLANK - this book's own 08-26 precedent is 'ADR-0012 no-guess blocks it'; (ii) tick (b) spans a TWO-session gap (08-27 missing); (iii) Jackson Hole = the book's largest single-event exposure at +/-3.5% = +/-2.27 ATR on the heaviest line, stance tighten-trail, catalyst-aware rule forbids pre-trim absent a confirmed lower-high or stop break - neither present; (iv) overlay override still says hold unless below recorded stop. Stop HELD 159.50, model 157.03 NOT adopted (lower). RE-ARM: execute the 50% trim on a close < 159.50, or re-test the fire at the 2026-09-01 EOD. |
| 0823 | LINK REIT | 39.26 | HOLD | — | 38.70 | n/a | n/a | n/a | Holding Link REIT into the Fed Chair's speech. The trim trigger fired, but on the same faulty measure affecting the rest of the book, and a property trust is about as rate-sensitive as a holding gets - acting the session before a major rates event is guessing, not managing risk. The stop stays where it is. Ops: FIRED armed + 2 ticks (c) T12W 38.89 < 40.05, (d) 2 consecutive NO TRADE (T5,T6) -> would be 50%. Deferred: same T12-progress arm artifact; Jackson Hole +/-3.5% tighten-trail plus HK rates data 09-02. Stop HELD 38.70 (chandelier 2.0xATR = 38.24 is LOWER, not adopted). Pillars Conflicted. Re-arm on close < 38.70. |
| 2888 | STANCHART | 230.40 | HOLD | — | 229.36 | n/a | n/a | n/a | Holding Standard Chartered. The position is already on a hair trigger - the stop sits less than half a percent below the current price - so there is no case for a discretionary trim on top of that, particularly the session before a major central-bank event that will reprice every bank in the book. Ops: armed + 1 tick (c) T12W 221.14 < 235.01 -> would be a 30% trim ~HKDeferred (Jackson Hole, HK banks). Stop HELD 229.36 vs close 230.40 = +0.45% - effectively pre-triggered. Model 220.27 NOT adopted (lower). |
| 0981 | SMIC | 70.15 | HOLD | — | 69.63 | n/a | n/a | n/a | Holding SMIC with the stop unchanged and close underneath. Korea's monthly semiconductor export data - the earliest hard read on chip demand - lands next week, and that is the number worth waiting for. Ops: PROFIT_ARMED but NOT fired - no tick qualifies: cv is -16 (under the 20-pt bar), T12W 76.01 well above the 1.02 threshold, no NO TRADE prints. Mandatory trail-raise only per ADR-0012. Stop HELD 69.63 vs close 70.15 = +0.75%, model 63.18 NOT adopted (lower). 09-01 Korea semi exports stance hold-through. |
| 0700 | TENCENT | 455.20 | HOLD | — | 431.48 | n/a | n/a | n/a | Tencent continues to flash a strong sell on every reading, but we already made this decision: the exit was escalated once, not filled, and retired to a standing trigger at HKThe shares are almost 6% above that line. Re-issuing the same sell every day is precisely the pattern that cost this book real money in July, so the line stands and the stop moves up instead. Ops: STABLE STRONG SELL - REDUCE on all 6 snapshots, Sell_Rank 1, cv - would otherwise be a top DISPOSE at HKNOT re-issued: HK-11 ('never re-issued' once retired) plus the live overlay override (trim-not-exit, keep 50% core, stop ~430, invalidated only on a close below 430). Close 455.20 = +5.9% above the line. Stop RAISED 430.00 -> 431.48 (model stop is HIGHER than the standing line, so adopted). |
| 7709 | XL2CSOPHYNIX | 36.44 | HOLD | — | 28.00 | n/a | n/a | n/a | Holding the leveraged Hynix ETF on its existing trigger at HKThe sell reading here is weak, and this instrument has a track record of bouncing hard after exactly this kind of low-conviction signal, so the standing rule is to trim rather than dump - and neither condition for that is met today. Ops: STRONG SELL - REDUCE on T4/T5/T6 but at cv 37. Retired WATCH 28.00 stands (HK-11); close 36.44 = +31% above. Leveraged-2x-ETF rule independently blocks a full exit (needs conv >= 70 + 3 prints + price < stop). Stop HELD 28.00. Overlay's 'stop 101' override is obsolete against a 36.44 close and was ignored. |
| 1347 | Hua Hong Semiconductor | 120.60 | HOLD | — | 105.00 | n/a | n/a | n/a | Holding Hua Hong on its existing HKtrigger; the shares are at 120.60 and rose 2.6% over the gap. A decision on whether the stock joins the Hang Seng Index is due in early September, which is worth waiting for. Ops: AVOID x5, no sell label, so nothing fresh to act on. Model printed Stop_Price 120.60 = THE CLOSE, a zero-width stop - rejected outright under HK-12 ('a stop that can never bind is not a risk control'). Stop HELD 105.00. 09-07 Hang Seng Index membership decision pending. |
Outcome & track record
Accuracy at the time of this record.
184 Graded decisions Past recommendations old enough to score against what actually happened (target hit / stopped out / closed). This is the evidence base — the bigger the number, the more history behind the calls. Click to inspect every one. See the decisions →
-0.261 1-week calibration How much of the eventual 4-week move typically shows up in the first week (0 = almost none, 1 = most of it). Use it to read the 1-week estimates: a low number means prices drift toward the target slowly, so don't over-read a soft week-1.
49 Published records How many dated end-of-day verdicts exist for this book — the length of the public track record you can browse in the Archive below.