Hong Kong — End-of-Day Verdict
2 ACQUIRE (0992 BUY 2000 @25.16 market; 0823 ADD 1700 QUEUED buy-limit @38.82 exp 08-07) + 3 DISPOSE (2513 SELL 400/40% @1010; 0981 SELL 1000/40% @65.30; 0005 SELL 3600/30% @166.50). 0700 and 3690 queues WITHDRAWN. 0939 (2 ticks) + 0823 + 2888 profit-take arms declined/displaced, ledgered optional. Stops raised on 0522 (137.00), 3750 (611.00), 2888 (224.83), 0981 remainder (58.14). No stop breaches. ANTICIPATE slot empty.
Recommendations
2 to acquire · 3 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 |
|---|---|---|---|---|---|---|---|---|---|
| 0992 | Lenovo Group Limited | 25.16 | BUY | 25.16 | 23.76 | 25.39 (+0.9%) | 26.15 (+3.9%) | 27.27 (+8.4%) | Lenovo is the one new-entry candidate in the Hong Kong book that is both technically triggered and fundamentally clean: the signal has printed BUY on five consecutive scans with conviction rising, the shares sit exactly at the model entry rather than extended above it, and the fair-value screen puts the stock at roughly fair with a strong quality score (F-Score 7) - the only name in the book to pass that gate. Valuation trims a little enthusiasm and the model's track record on this signal family argues for caution, but the combination of a rising trigger, an unchased entry and passing quality is the best buy setup available today. Sized at the 2,000-share board-lot floor (a 41% round-down from the model's HK) because the stock is up ~115% in 30 days and swung 5.9% intraday - the round-down IS the de-sizing. Ex-dividend HKon 2026-08-05 (the execution session): expect a mechanical ~1.3% markdown to ~24.82 at the open, still inside the 24.76-25.57 entry band; do not misread it as a signal break. Adjusted conviction 83 -2 val -4 bias = 77. Entry range 24.76-25.57. Learnings: HK-1 confirmation requirement met (5 prints, RSI 58.9 < 70, eD 0.00); L8 pending-next-open. · news |
| 0823 | LINK REIT | 39.02 | ADD — QUEUED @38.82 | 38.82 | 37.84 | 38.93 (+0.3%) | 39.13 (+0.8%) | 39.20 (+1.0%) | Link REIT has pulled back to its volume-weighted average price inside an intact uptrend - the single most reliable buy setup this book has produced, with three of every four past instances hitting their one-week objective. Conviction has risen across three consecutive scans, the shares trade essentially at fair value, and a defensive REIT is a sensible place to redeploy part of the cash raised from today's semiconductor and bank disposals. The order is a limit 0.5% below the close so we buy the dip rather than chase it. Honest limitation: every modelled horizon is +1.0% or less - this is a low-volatility carry add, not a return driver, with roughly HKof modelled 4-week gain against HKof stop risk (R:R ~1.3 to the 40.33 objective). Adjusted conviction 67 +0 val +5 bias = 72. Size 63.97% x 2,800 = -lot floor -> = HKEntry range 38.68-38.96. Quality gate F=6/False is NON-BINDING here - it blocks new entries only and 0823 is held. Position stop stays 37.84 (the model's 37.68 would lower the recorded line - refused per trail discipline). Expires 2026-08-07 (anchor + 3 sessions). The same row is PROFIT_ARMED with 1 live tick - arm DECLINED because conviction rose on the book's best-graded buy cell (G3: do not cut a winner while the signal is bullish). Learnings: L1/HK-2 prefer the PullbackBuy anchor; HK-4/HK-10 no-chase strict. |
| 2513 | KNOWLEDGE ATLAS (Z.AI) | 1010.00 | SELL 40% | 1010.00 | 862.00 | 1050.26 (+4.0%) | 1011.36 (+0.1%) | 1112.15 (+10.1%) | Z.AI jumped 7.5% today and the model used that strength to escalate from a passive avoid to its strongest sell rating - a failed-breakout pattern that has been the book's best-evidenced exit signal across 48 past instances. The rally has not repaired anything structurally: the shares remain far below both their 20-day and 50-day trend lines, and the company has just launched an placement, which dilutes existing holders. Selling into strength on the book's second-largest position, which is down 38% from cost, takes risk off at a materially better price than was available yesterday. We are deliberately keeping 60% (, HK) because the Hang Seng quarterly index review on 2026-08-14 names Z.AI as an inclusion contender with an asymmetric -6%/+12% band, so this is a de-risk, not an exit. Proceeds HKDEVIATION from the 08-03 WATCH line (re-arm <862): that trigger governed a downside re-arm of a RETIRED card; HK-11 forbids re-issuing a retired card, not acting on a NEW signal-based origination, and 1,010 is strictly better than the 940 the WATCH was set at. HK-9 caps a signal-only sell to a tactical 1-week de-risk -> 40%, not the engine's 85.46%. Estimator contradicts at all three horizons (+4.0/+0.1/+10.1%) - systemic finding #2; the 40% portion IS the concession. HK-12: printed Stop_Price 629.61 on a 1,010 close (-37.7%) is degenerate width, rejected; remaining keep the manual 862.00 line. · news |
| 0981 | SMIC | 65.30 | SELL 40% | 65.30 | 58.14 | 66.91 (+2.5%) | 70.83 (+8.5%) | 72.41 (+10.9%) | SMIC has now printed the model's strongest sell rating on five consecutive scans, with conviction rising into today's 5% bounce - the classic sell-the-rally setup this exit signal is graded on, and one that has worked about two-thirds of the time over 48 past instances. Despite the rally the shares remain below their 20-day, 50-day AND 200-day trend lines, so the downtrend is intact, and the stock is expensively valued against our fair-value anchor. Trimming 40% ahead of Q2 results on 2026-08-06 cuts exposure to a two-sided event while leaving for a beat. Proceeds HKSUPERSEDES and reprices the unexecuted 08-03 card @62.20 - the one-session delay earned +HKValuation amplify AVAILABLE (MoS -4.39 <= -3 would bump a tier) and DECLINED per HK-9, which caps a signal-only sell to a tactical 1-week de-risk. HONEST COUNTER-ARGUMENT on the record: the Event Horizon stance for the 08-06 print is hold-through (+-9%), not tighten-trail, and the estimator contradicts the signal at every horizon (+2.5/+8.5/+10.9%) - systemic finding #2. Trading the 48 graded outcomes over the estimator, sized down to 40% because of the disagreement. Remaining stop raised 58.08 -> 58.14 (1.5x ATR). |
| 0005 | HSBC HDGS(USD0.50) HK O'SEAS | 166.50 | SELL 30% | 166.50 | 163.17 | 165.94 (-0.3%) | 165.26 (-0.7%) | 169.53 (+1.8%) | HSBC delivered strong first-half results today - pre-tax profit up 23% to 5bn, a resumed $1bn buyback and a second interim dividend - and the London line rose on the news, but the Hong Kong shares sold it: a higher high at 169.70, then a lower low at 164.00, closing down 1% at 166.50 in the lower third of the day's range. That is a classic exhaustion reversal on a stock that has run vertically to an RSI of 73, and our model now projects flat-to-negative returns at every horizon (-0.3% / -0.7% / +1.8%). Banking a third of the book's largest position at the top of that run is prudent profit-taking, not a change of view - we keep and the full participation in the improving fundamentals. Proceeds HKon a HKposition. ADR-0012 profit-take: ARMED (PROFIT_ARMED, T12_Progress 98.21) + 1 tick (c) dead forward view (T12 169.53 < 1.02 x close 169.83) -> profit_take_1 = 30%. Ticks (b)/(d) explicitly do NOT fire: NO TRADE prints cv=0 as a sentinel and must not be read as a -45pt collapse, and 08-03's settled close was HOLD 45 so there are not 2 consecutive NO TRADE EOD prints. NEWS OVERRIDE DEMOTES the unexecuted 08-03 50% card to 30% - a 23% profit beat with a resumed buyback is not deterioration, so the card is demoted one tier rather than carried forward or cancelled. 3,600 / 400-lot = exact. Valuation NA_CCY (MoS -11.16 unusable) -> no amplify. The 08-13 ex-div is forgone on the sold 30% (~HKimmaterial). Remaining 8,400 trail 163.17 unchanged (1.7x ATR computes to 160.53 and would lower the line - refused). · news |
| 0522 | ASMPT | 158.00 | HOLD | 158.00 | 137.00 | 160.06 (+1.3%) | 175.49 (+11.1%) | 173.60 (+9.9%) | ASMPT rallied 5.1% with the rest of the semiconductor complex and the model has moved off its bearish reading to neutral. No action warranted either way, but with the shares now 19% above the protective stop we are raising that line to lock in more of the rebound. Still 19.5% below cost. Stop RAISED 133.00 -> 137.00 (1.5x ATR from the 158.00 close). MoS -3.92 would amplify a sell one tier, but there is no signal-based sell to amplify and valuation never originates one (CLAUDE.md section 9). The 07-31 advisory DISPOSE remains void and is not re-issued (HK-11). |
| 1347 | Hua Hong Semiconductor | 135.10 | HOLD | 135.10 | 118.00 | 137.18 (+1.5%) | 139.23 (+3.1%) | 137.69 (+1.9%) | Hua Hong jumped 6.6% in the semiconductor rebound and the model has upgraded from bearish to neutral. Q2 results land on 2026-08-06 and the event guidance is to hold through them, so no action - the position sits 14.5% above its protective stop with a modestly positive forward view. Stop HELD at 118.00 (a 1.5x ATR trail from 135.10 computes to 114.12 and would lower the line). L2 exit test not met: no stop break and T12 137.69 is +1.9% positive. |
| 3750 | CATL | 649.50 | HOLD | 649.50 | 611.00 | 649.63 (+0.0%) | 633.58 (-2.5%) | 614.61 (-5.4%) | CATL rose 3.9% today but is the one holding whose signal got WORSE on an up day - the model moved from neutral to a bearish avoid, and its 12-week projection (614.61) now sits 5.4% BELOW the current price. That is not yet an actionable sell (an avoid rating is not a sell rating, and the shares are still 12% above the stop), so the response is a materially tighter trailing stop rather than a trade. Re-check next session. Stop RAISED 580.00 -> 611.00 (1.5x ATR from the 649.50 close). L2 exit test: negative 12wk leg IS met but the stop-break leg is NOT (649.50 vs 580.00), so no card. Escalated NO TRADE cv0 x4 -> AVOID cv63. Position -13.61% vs cost 751.80. |
| 7709 | XL2CSOPHYNIX | 37.10 | HOLD | 37.10 | 33.00 | 39.54 (+6.6%) | 37.31 (+0.6%) | 38.49 (+3.7%) | This leveraged ETF bounced 4.7% with the Korean/semiconductor complex but remains down 68% from cost. The rating is a bearish avoid, not a sell, and the leveraged-ETF rule requires a much stronger case (high conviction, three consecutive prints and a price below the stop) before cutting - a rule written precisely because a previous panic exit here missed a 5.8% bounce. Holding against a hard risk line at 33.00. No card: not an actionable sell label; L2 not met (no stop break, T12 38.49 is +3.7%). HK-11 retired its card. Risk line 33.00 is explicitly a hard CAP, not an ATR stop - ATR 21.65 on a 37.10 close is 58% of price and the printed Stop_Price 80.40 sits 117% ABOVE the close (HK-12 degenerate, rejected). |
| — | Sunny Optical Technology | — | NO ACTION | — | — | n/a | n/a | n/a | Sunny Optical is the only name in the entire Hong Kong universe that screens as genuinely undervalued AND passes the quality test - worth about 20% more than it trades for, on a 7/9 fundamental score. But the price action disagrees: the model rates it a bearish avoid, and buying a cheap stock in a confirmed downtrend is the falling-knife trade this book has repeatedly been punished for. Logged as a watch candidate, not a buy. Valuation NEVER originates a buy on its own - it needs a bullish technical trigger, which is absent (AVOID cv76). Re-evaluate the moment it prints a BUY-side signal clearing the 65 floor. Book-specific note B: the HK buy pool is persistently oversold/reversal names with negative 12wk views - not buyable. |
Outcome & track record
Accuracy at the time of this record.
136 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.