Hong Kong — End-of-Day Verdict
2 ACQUIRE (0823 ADD 1700 QUEUED buy-limit @38.94 exp 08-10; 0992 BUY 2000 QUEUED buy-stop @26.63 exp 08-10) + 3 DISPOSE (0005 SELL 6000/50% @162.10 market - STOP BREACH + profit-take 2 ticks; 2513 SELL 400/40% @1041 market; 0522 SELL 550/50% @165.60 market, valuation-amplified one tier). 0981 RETIRED to WATCH per HK-11. 7709 (adj 73) clears the floor but is displaced by the <=3 cap. 0939 (2 ticks), 2888 (1 tick) and 0823 profit-take arms declined/displaced, ledgered optional. 0700 add falls below the floor at adj 60; 0388 BLOCKED by the quality gate (F=4) and adj 55. Stops: 0005 re-based 159.50, 0522 raised 145.50, 0981 raised 58.86, 0700 raised 466.94, 3750 raised 625.50; 1347 held 118.00 into its Q2 by design. 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 |
|---|---|---|---|---|---|---|---|---|---|
| 0823 | LINK REIT | 39.06 | ADD — QUEUED @38.94 | 38.94 | 37.84 | 39.04 (+0.3%) | 39.18 (+0.6%) | 39.93 (+2.5%) | Hong Kong's largest REIT pulled back to its volume-weighted average price and the buy signal strengthened sharply into the close, taking conviction from - and this pullback-to-VWAP pattern is the single most reliable buy setup this portfolio has traded, hitting its one-week objective in 3 of the last 4 cases. The shares sit essentially at our estimate of fair value, so we are paying a fair price for a steady income asset rather than chasing a rally. Adding 1,700 units at HKon a limit order, keeping the risk line at HK. Honest caveat: every forward estimate is under +2.5%, so this is a low-volatility carry add, not a return driver - the modelled 4-week gain of roughly HKsits against HKof stop risk. Operational: GTC buy-limit, expires 2026-08-10, entry range 38.80-39.08; adjusted conviction 81 = 76 + 0 val + 5 bias, the highest in the book today. Quality gate F=6 fails but blocks NEW entries only - 0823 is held, so it does not bind (stated, not ignored). Same row prints PROFIT_ARMED; arm DECLINED per G3 (conviction rose 9 pts on a bullish signal - do not cut a winner). Model stop 37.79 refused, would lower the recorded 37.84. 12wk 39.93 exceeds the DCF anchor 37.84 - projection assumes growth beyond priced-in. FILL AMBIGUITY: the 08-04 GTC @38.82 was touched intraday (low 38.64) but the book is unchanged - if that order actually filled, mark this card SKIPPED as a duplicate. L1/HK-2 positive anchor; L8 pending_next_open. |
| 0992 | Lenovo Group Limited | 26.28 | BUY — QUEUED @26.63 | 26.63 | 22.46 | 26.78 (+0.6%) | 27.25 (+2.3%) | 27.14 (+1.9%) | The world's largest PC maker has broken out to record highs on an AI-server earnings beat - its infrastructure division grew AI server and datacentre revenue 37% - and it is the only new-entry candidate in this portfolio that passes our quality screen for balance-sheet strength and cash generation (F-Score). The shares trade near our assessment of fair value despite the run, and a FIFA World Cup 2026 technology partnership keeps the story in front of investors. We are buying only on further strength, via a stop order at HKabove the current price, so the position opens only if the breakout continues. The real risk here is extension: the stock is up roughly 142% in 90 days and our own forward estimates are flat-to-+2% at every horizon against a +31% price target - the estimator does not believe the target. Operational: GTC buy-stop, expires 2026-08-10, entry range 26.22-27.04; adjusted conviction 80 = 85 - 2 val - 3 bias. Sizing 71.55% x 600k x 20% = HK -> -> 2,000-share board lot = HK (38% round-down; do NOT round to 4,000). max_position_pct 60% not binding. HK-1 confirmation MET (4 consecutive rising buy prints, RSI 62.6 < 70, entry above close = no chase); BreakoutUp_20D|buy cell improved -5 to -3 this run. L6 (no post-earnings gap-chase) is the standing objection, answered by the buy-stop structure + lot-floor de-sizing, not dismissed. Third consecutive session carded unexecuted - this is its HK-11 escalation print; if unexecuted again 0992 RETIRES to a WATCH line. Card walked 24.74 -> 25.16 -> 26.63 (+7.6%); repriced to live Entry_Price per provenance. 12wk 27.14 below DCF anchor 43.47 - no flag. L8 pending_next_open. · news |
| 0005 | HSBC HDGS(USD0.50) HK O'SEAS | 162.10 | SELL 50% | 162.10 | 159.50 | 161.00 (-0.7%) | 160.57 (-0.9%) | 164.88 (+1.7%) | HSBC broke its risk line today. The shares fell 2.6% to close at the session low of HK, below our HKstop and below the 10-day average, and the signal flipped to a sell warning citing confirmed exhaustion after a sustained overbought run. What makes this decisive is the context: the bank reported a strong first half only one session ago - pre-tax profit up 23%, buybacks resumed, dividend raised - and the market has sold that beat on both sessions since, with a lower high and a close at the low. A result that good being distributed twice is a distribution signal, not support. We are taking half the position off at roughly HKand keeping half, with the remaining risk line reset to HK. Operational: THREE stacked legs - (1) stop breach 162.10 vs 163.17, closed at the day's low 162.00 after a lower high; (2) label NO TRADE -> SELL - WARNING, Portion 75.38, Sell_Rank 5, below EMA10 163.59; (3) ADR-0012 profit-take FIRED with 2 ticks - (a) EMA10_BREAK, a Tier-2 code, and (c) dead 12wk view (164.88 < 1.02 x 162.10 = 165.34) -> profit_take_2 = 50%, not the 30% tier. G1 inapplicable (no fresh breakout >=80). 6,000/400-lot = exact. Proceeds HK - the largest banked-dollar action in the book. Estimator AGREES here (-0.68%/-0.94% at 1wk/4wk), unlike the semis rows. Valuation NA_CCY, no amplify. HK-11 session 3 -> market-at-open escalation, ONCE; if unexecuted by the 08-06 close, 0005 retires to a WATCH line and is not carded again. 08-13 ex-div /sh forgone on the sold half ~HKimmaterial. Model stop 129.68 (-20%) rejected per HK-12; 159.50 sits just under EMA20 160.38 and well above EMA50 153.45 - a re-base AFTER a triggered breach that produced a trim, not a ratchet-down of an unbreached line. L8 pending_next_open. · news |
| 2513 | Z.AI Co., Ltd. (ex-KNOWLEDGE ATLAS) | 1041.00 | SELL 40% | 1041.00 | 862.00 | 1069.83 (+2.8%) | 1041.82 (+0.1%) | 1157.36 (+11.2%) | The Chinese AI model developer bounced 3% today but could not even reclaim its 10-day average, and the sell signal has now printed for a sixth consecutive session on the pattern that has been this portfolio's most reliable exit trigger across 50 graded trades. The company's roughly US$4bn share sale was priced at HK-1,698 - between 53% and 63% above where the stock closed today - which means the investors who funded that raise are deeply underwater and represent an overhang rather than a floor. A company that raised at a price the market has since repudiated by half is not a bargain yet. We are selling 40% at about HKand keeping into the 14 August Hang Seng index review, where inclusion would be asymmetrically positive. Operational: sixth consecutive STRONG SELL - REDUCE cv73, Sell_Rank 3; adjusted 76 = 73 + 0 val + 3 bias. Close 1,041 vs EMA10 1,070.66 vs EMA20 1,212.52 vs EMA50 1,312.86; today's 1,110 high rejected by 69 pts into the close, RSI 42.9. Engine portion 85.46% capped to 40% per HK-9 (sell signals are a 1-week instrument) - deliberately leaving (HK) exposed to the 08-14 index review, band -6%/+12%, confidence LOW, 7 sessions out (outside the <=5-session catalyst hold, so it sizes the trim rather than blocking it). Position -36.33% vs cost 1,635; proceeds HKon the book's 2nd-largest holding. WARNING: estimator contradicts at every horizon (+2.8/+0.1/+11.2%) - systemic finding #2, fourth consecutive run; trading the 50 graded outcomes over the estimator, with the 40% portion as the concession. HK-11 session 2, inside the escalation allowance, so repricing from the 08-04 @1,010 card is legitimate. Printed Stop_Price 1,040.99 = the close to one cent - degenerate, rejected per HK-12; remaining 600 keep the manual 862.00. L8 pending_next_open. · news |
| 0522 | ASMPT | 165.60 | SELL 50% | 165.60 | 145.50 | 167.98 (+1.4%) | 177.88 (+7.4%) | 187.62 (+13.3%) | This semiconductor assembly-equipment maker rallied 4.8% today but stalled right at the underside of its 20-day average and remains below its 50-day, and the signal escalated into that bounce - it is now the strongest sell reading in the whole portfolio, confirmed across four consecutive intraday prints rather than a single flip. The shares also screen as expensively valued against what the business earns, which is why we are taking half rather than a third. The position is 15.7% underwater against cost, so this is cutting a losing holding into strength rather than hoping the bounce continues. Selling at roughly HKand keeping 550 with the risk line lifted to HK. Operational: NEW entrant to the dispose pool at the TOP of the conviction column - STRONG SELL - REDUCE cv81, Sell_Rank 1, adjusted 84 = 81 + 3 bias; trail -> across four intraday prints, so a confirmed state change, not a single-print flip. Close 165.60 vs EMA10 156.67 (reclaimed), EMA20 164.22 (only just), EMA50 171.30 (still below); RSI 48.7; 168.80 high rejected by 3.2 pts into the close. VALUATION AMPLIFY FIRES: MoS -3.92 <= -3 bumps HK-9's tactical 40% one tier to 50% (CLAUDE.md 9b) - amplification of a signal-originated sell, never origination. = 100-lot exact = HK; position HK, -15.66% vs cost 196.35. WARNING: estimator contradicts at every horizon (+1.4/+7.4/+13.3%) - systemic finding #2. Applied Materials FQ3 on 08-13 (semicap read-through, +-4%, hold-through) is 6 sessions out, outside the <=5-session catalyst hold, so it does not block. Ex-div HKon 08-11 forgone on the sold half ~HKimmaterial. Printed Stop_Price 165.60 = the close EXACTLY - degenerate, rejected per HK-12; remaining 550 get 145.50 (1.5 x ATR 13.30 from the close), a raise from 137.00. L8 pending_next_open. · news |
| 0981 | SMIC | 68.00 | NO ACTION | 68.00 | 58.86 | 68.71 (+1.0%) | 72.72 (+6.9%) | 73.08 (+7.5%) | China's largest foundry still carries a valid sell signal - a sixth consecutive reading - but it reports second-quarter results tomorrow, and a set of numbers that can move the shares 9% either way is not something to pre-position against on a signal that has already been acted on twice without execution. The stock has also climbed back to its 200-day average, so the structural picture is genuinely less one-sided than it was a week ago. We are standing down rather than issuing the same trim a fourth time, and will re-engage only if the shares break lower after the results. Operational: NOT carded for three converging reasons - (i) HK-11: carded 08-03 and 08-04, both already market-at-open, so the one-time escalation is SPENT and a fourth issue is exactly the behaviour HK-11 exists to stop; (ii) displaced on dollar bleed by 0522 (HK vs HK) for the last of three slots; (iii) Q2 prints 08-06 at a +-9% two-sided band with a horizon stance of hold-through. RETIRED to a WATCH line with a NUMERIC re-arm: a settled close below 65.50 (today's low) AFTER the 08-06 Q2 print, OR any settled close below the 58.86 stop. Close 68.00 below EMA20 69.97 and EMA50 72.55 but has reclaimed EMA200 68.65 to within 1%. Valuation amplify (MoS -4.39 <= -3) available and DECLINED with the card. Stop raised 58.14 -> 58.86 (the model line, higher than recorded). Ledgered optional so the counterfactual grader prices what not acting costs. The 08-04 delay earned +HKfor the second session running. |
| 7709 | XL2CSOPHYNIX (2x Hynix ETF) | 38.94 | NO ACTION | 38.94 | 33.00 | 40.79 (+4.8%) | 38.58 (-0.9%) | 40.32 (+3.5%) | This 2x leveraged Hynix tracker printed its first genuine sell signal in weeks, clearing the conviction bar after four consecutive readings, and a 50% trim would be the right response. It loses its slot only because at roughly HKit is the smallest of the five sell candidates today and three larger, more damaging positions rank ahead of it. Worth noting the rule that applies here: leveraged products get trimmed, never dumped wholesale, because they bounce as violently as they fall. Operational: AVOID cv62 -> STRONG SELL - REDUCE cv70, Sell_Rank 4, four consecutive prints; adjusted 73 = 70 + 3 bias - it CLEARS the floor and is displaced purely by the <=3 dispose cap. Close 38.94 (+4.96%), -66.01% vs cost 114.56, position HKLeveraged-ETF rule satisfied on conv >=70 and >=3 prints but NOT on price < Stop_Price: printed stop 2.81 is HK-12 degenerate (-93% below close) and the manual hard cap 33.00 is intact at 38.94 - so even with a slot the action would be a 50% trim, NEVER a full exit. Hard risk cap 33.00 RETAINED and is explicitly a cap, not an ATR stop (ATR 18.06 on a 38.94 close = 46% of price). Ledgered optional so the counterfactual grader prices the non-action. |
| 0939 | CCB - H SHS | 8.80 | NO ACTION | 8.80 | 8.78 | 8.79 (-0.1%) | 8.75 (-0.6%) | 8.58 (-2.5%) | The mechanical profit-take trigger fired on this mainland bank for a third straight session and we are declining it again, because the trigger is measuring the wrong thing: it arms whenever the 12-week outlook drifts toward the current price, which happens on losing positions as readily as winning ones, and this holding has no recorded to verify a profit at all. A 2.5% drift in the 12-week view on a low-beta bank is noise, not deterioration. The line to watch instead is the stop at HKagainst a HKclose - the tightest in the portfolio, where a single 0.3% down-day converts this into a real exit case. Operational: mechanically armed (T12_Progress 102.56) with 2 ticks - (c) T12 8.58 < 1.02 x 8.80 = 8.98, and (d) NO TRADE on both the 08-04 and 08-05 settled closes - which would mandate a 50% trim = = HKthe largest banked-dollar candidate after 0005. DECLINED on unchanged grounds: PROFIT_ARMED prints on held names today including 7709 at -66.01% and 2513 at -36.33%, all with negative Gain_ATR; Avg_Cost blank so the profit limb is unverifiable; 'Consensus conflict + low conviction' is 0939's permanent state. THIRD consecutive decline - counterfactual pressure noted (optional_trim wh_1w +2.78% 'saved', though the wh_4w figure swung 47 pts on a 2-sample increment and cannot carry weight). Stop HELD at 8.78, not raised: 1.5 x ATR from 8.80 computes to 8.45, lower. Top candidate for the next hk learnings review alongside systemic finding #1. |
| 2888 | STANCHART | 234.00 | NO ACTION | 234.00 | 224.83 | 235.64 (+0.7%) | 234.09 (+0.0%) | 232.22 (-0.8%) | Standard Chartered is a genuine winner - 23.5% above its 200-day average and holding up while HSBC broke down today - and the only reason it appears here at all is a mild profit-take trigger from a flat 12-week outlook. At roughly HKa 30% trim is the smallest action available in the portfolio and is displaced by three far larger cases. Holding, with the trailing stop unchanged at HK. Operational: armed + 1 tick - (c) T12 232.22 < 1.02 x 234.00 = 238.68. 30% tier = -> 50-lot -> = HKthe smallest banked-dollar candidate in the book, so 0005/2513/0522 displace it on the <=3 cap. Close 234.00 vs EMA50 217.38 vs EMA200 189.40 (+23.5%), RSI 63.3. Est 1wk 235.64 (+0.7%), 4wk 234.09 (0.0%), 12wk 232.22 (-0.8%) - flat, not deteriorating. Stop 224.83 HELD (1.5 x ATR computes to 223.89, lower). Ledgered optional for the counterfactual grader. |
| 0700 | TENCENT | 492.20 | HOLD | 492.20 | 466.94 | 487.27 (-1.0%) | 492.93 (+0.1%) | 518.26 (+5.3%) | Tencent's signal turned constructive again and the engine wants to add, but after adjusting for a full valuation and the weak track record of breakout entries in this portfolio the case falls just short of our threshold - and the shares still trade below their 200-day average, meaning the primary downtrend has not been repaired. Second-quarter results land on 12 August and are the single largest event risk in the portfolio over the next fortnight, capable of moving the shares 7% either way. Holding the existing position, lifting the risk line to HK, and not adding into the print. Operational: BUY - ADD cv66 -> adjusted 60 = 66 - 3 val - 3 bias, BELOW the 65 floor. Engine wanted a 64% add via buy-stop @498.51 - declined. Close 492.20 still below EMA200 504.17; RSI_DIV also prints. Stop RAISED 466.06 -> 466.94 (the model line). Tencent Q2 08-12 is +-7% = +-2.03 ATR, WIDER than a 1.5-ATR stop, so a bad print alone can carry the position through its stop in one session without the signal degrading first; horizon stance tighten-trail. Do not add into it (L6). Re-assess before 08-12. |
| 0388 | HKEX | 414.60 | WATCH | 412.43 | 397.06 | 414.90 (+0.6%) | 414.11 (+0.4%) | 413.39 (+0.2%) | The Hong Kong exchange operator printed a fresh breakout buy signal today, but it fails our quality screen for balance-sheet strength and cash generation outright, and it screens as expensive against what the business earns - so it is blocked as a new entry on both counts. The shares also closed at the day's high, meaning any purchase now would be chasing. Not a near-miss: a clean rejection. Operational: BreakoutUp_20D BUY cv64, Buy_Rank 4, close 414.60 (+1.62%) at the session high, RSI 69.5. TWO independent blocks - (i) Quality Gate: Quality_Pass False, F-Score 4, hard-blocks NEW entries; (ii) adjusted 55 = 64 - 6 val - 3 bias, ten points under the 65 floor, OVERVALUED_VS_FV MoS -1.12. Also eD -0.36 breaches the no-chase bar (<= -0.25) with the close at the high. No card. |
| 1347 | Hua Hong Semiconductor | 141.50 | HOLD | 141.50 | 118.00 | 141.33 (-0.1%) | 146.56 (+3.6%) | 150.09 (+6.1%) | This foundry rallied 4.7% today and reports second-quarter results tomorrow with the potential to move 10% either way. There is no sell signal, the 12-week outlook is still positive, and the position sits comfortably above its risk line - so we sit through the print rather than tightening into it. Deliberately leaving the stop where it is: pulling it 3.3% closer ahead of a two-sided binary would invite a stop-out on noise rather than on a real breakdown. Operational: NO TRADE cv0, close 141.50 (+4.74%) vs stop 118.00 (+19.9% headroom), T12 150.09 = +6.1% POSITIVE, so L2 is not met on either leg. Horizon stance hold-through, band +-10%. Stop HELD at 118.00 deliberately - a 1.5 x ATR trail computes to 121.29 and would tighten into the print. |
| 3750 | CATL | 663.00 | HOLD | 663.00 | 625.50 | 663.98 (+0.1%) | 648.02 (-2.3%) | 632.12 (-4.7%) | The battery maker rose 2.1% today but its signal weakened for a second straight session and the 12-week outlook has now dropped below the current price - a deteriorating forward view on a rising share price. There is no actionable sell label and the position is nowhere near its risk line, so no trade, but we are lifting the trailing stop to HKto bank the move. Of everything held, this is the name most likely to produce a genuine exit case next. Operational: AVOID cv63 -> AVOID cv0 (NoSignal), second consecutive session of signal deterioration on an up day. Target_12Week 632.12 is BELOW the 663.00 close (-4.7%). AVOID is not an actionable sell label and L2's stop-break leg is not met, so no card. Stop RAISED 611.00 -> 625.50 (1.5 x ATR 24.96 from the close). Re-check next session. |
Outcome & track record
Accuracy at the time of this record.
140 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.