Hong Kong — End-of-Day Verdict
1 ACQUIRE (0992 Lenovo BUY 2,000 QUEUED buy-limit @27.32 exp 08-12, adj 71 - quality rank 1 in the book, sized down one lot per HK-1) + 3 DISPOSE (0939 CCB SELL 25,000/49% @8.79 market - ADR-0012 arm+fire, conviction -24 and dead 12wk, largest banked trim at HK; 0981 SMIC SELL 2,000/80% @66.90 market - stable STRONG SELL cv73 x4, Sell_Rank 1, sub-EMA200, adj 76, the book's highest; 3750 CATL SELL 200/67% @632.50 market - three ticks fired, forward view negative on all three horizons, 1.1% off its stop). Sell proceeds HKvs HKof conditional buying -; goes to ~HK. 0316 OOIL BLOCKED by the quality gate at adj 81, the highest reading of the day - negative FCF, revenue -9.2%. 0941 China Mobile rejected at adj 62 + gate. 2318 GTC WITHDRAWN. ANTICIPATE slot empty (no BUY - ANTICIPATE print). 2513 stays retired to WATCH after a +14.6% session; 0700 and 0005 held through 12-Aug/13-Aug catalysts. SYSTEMIC: PROFIT_ARMED fires on held names for a second session - the ADR-0012 arm machinery is non-discriminating in this book, so the two largest cards were selected on ticks (b)/(c)/(d) plus a negative forward view, never on the arm alone.
Recommendations
1 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 | 28.10 | BUY — QUEUED @27.32 | 27.32 | 22.78 | 28.71 (+5.1%) | 29.16 (+6.7%) | 28.40 (+4.0%) | Lenovo is the highest-quality name the HK book can currently buy - F-Score 7, ROE 28%, positive free cash flow and the best quality rank in the universe - and it has just broken to a 52-week high on a sovereign-AI infrastructure win in Saudi Arabia, with the shares up ~115% over 30 days. Fair value reads it as roughly fairly priced (MoS -0.39), so we are paying up for momentum rather than buying a bargain; that is why this is a limit order below the market and not a market buy. OPERATIONAL: no-chase applies - Entry_Distance_ATR -0.43, RSI 67.2, closed at the session high 28.10 - so this is a GTC buy-limit at the live model entry 27.32 (range 26.87-27.77), a -2.8% pullback that sat inside today's 25.34-28.14 range. Sized DOWN to one 2,000-share board lot (HK) against a HKmodel because HK-1 forbids up-sizing any momentum-chase family, and the BreakoutUp_52W|buy scorecard cell is n=2, report-only, no bias credit. Prior 08-05 queue at 26.63 was withdrawn and is NOT carried forward - repriced to the live entry per the provenance rule. FQ1 earnings 13-Aug: the queue expires 2026-08-12, one session BEFORE the print, so the order is off the book by design (L6). Stop 22.78 (chandelier trail). Board lot 2,000 is an unattended-run assumption - verify. · news |
| 0939 | CCB - H SHS | 8.79 | SELL 49.3% | 8.79 | 8.34 | 8.74 (-0.6%) | 8.73 (-0.7%) | 8.58 (-2.4%) | Banking the book's largest un-cut winner while the forward view has gone flat. CCB's conviction collapsed in a single session and every model horizon now points slightly DOWN from here - 1-week -0.6%, 4-week -0.7%, 12-week -2.4% - so the position has stopped paying us to carry its full size. This is a half-trim, not an exit: stay long the dividend and any re-rating. OPERATIONAL: ADR-0012 profit-take, armed (PROFIT_ARMED, T12_Progress 102.5) with TWO ticks fired - (b) conviction -24 pts vs the 08-06 settled close, and (c) dead forward view, Target_12Week 8.58 below 1.02x the 8.79 close - which sets the 50% tier. Rounded to 25,000 ( of 1,000) = HKthe largest banked-dollar trim available. Valuation NA_CCY so no amplification. This arm was displaced by the <=3 cap on four consecutive sessions (08-03/04/05/06) and the 08-06 memory pre-committed it a slot today; HK-5 says flagged trims left untaken in this book tended to be right. Residual stop moved to the model line 8.34 (2.0x ATR) - the old 8.78 state stop sat HKunder the close and would fire on noise. · news |
| 0981 | SMIC | 66.90 | SELL 80% | 66.90 | 58.86 | 67.36 (+0.7%) | 68.46 (+2.3%) | 70.96 (+6.1%) | SMIC is the only outright sell label in the book and it hardened today rather than fading: STRONG SELL - REDUCE at conviction 73 on all four snapshots, rank 1 on the sell side, with the shares still below their 200-day average (66.90 vs 68.58) after a failed attempt to break out. The chip maker is priced for growth its cash generation does not support - free cash flow is negative and fair value reads it -4.4 margin of safety - and today's +2.4% bounce into a bearish structure is the better exit we did not get yesterday. Taking 80% off and leaving. OPERATIONAL: adjusted conviction 73 + 3 = 76, the book's highest; FailedBreakout_Down_20D|sell is the best-sampled cell here (n=55, hit_1w.646, bias +3). Model portion 85.46% x 2,500 = 2,136 rounded DOWN to 2,000 ( of 500) = HKValuation amplify SUPPRESSED by gate G2 - DCF_per_share is blank, so there is no reliable anchor to bump the tier from; portion is the model's, not an escalation of yesterday's 40%. The 08-06 card @65.35 went unexecuted; this is session 2 since the 08-05 re-arm, so HK-11 does not yet force escalation-and-retire. Model stop 58.78 today is sane (-12.1%); residual stop held at 58.86. All three model estimates lean AGAINST the exit (+0.7/+2.3/+6.1%) and are discounted per HK-6 (calib_1w -0.249) and HK-9 (HK sell signals are a one-week instrument). Read-through risk both ways: TSMC July revenue ~10-13 Aug, AMAT FQ3 13-Aug. · news |
| 3750 | CATL | 632.50 | SELL 66.7% | 632.50 | 625.50 | 630.10 (-0.4%) | 617.17 (-2.4%) | 596.19 (-5.7%) | The battery maker has gone quiet and the model's forward view has turned negative across every horizon - 1-week -0.4%, 4-week -2.4%, 12-week -5.7% - while the position sits just 1.1% above its stop and 15.9% underwater from a 751.80 cost. There is no live thesis left to carry three-quarters of the size on: the signal engine has printed NO TRADE all session and conviction went. Cutting to a 100-share token. OPERATIONAL: ADR-0012 profit-take, armed with THREE ticks fired - (b) conviction -52 pts, (c) Target_12Week 596.19 below 1.02x the 632.50 close, (d) repeated NO TRADE - which sets the 50% tier. 50% of 300 = 150, not a lot multiple; rounded UP to 200 ( of 100) = HKrather than down to 100, because a dead forward view 1.1% off the stop makes the overshoot the safer error. Valuation FAIR_VALUE, no amplification. Residual stop held tight at 625.50 (0.23x ATR under the close) by design on a deteriorating name. Board lot 100 is an unattended-run assumption - verify before placing. · news |
| 7709 | XL2CSOPHYNIX | 31.06 | NO ACTION | 31.06 | 28.00 | 32.63 (+5.1%) | 32.25 (+3.8%) | 32.79 (+5.6%) | The 08-06 SELL 900 @31.50 card was NOT executed and is NOT withdrawn - it is displaced by the <=3 dispose cap on size, ranking fourth at HKof banked value behind CCB, SMIC and CATL. The surface case did weaken: no stop breach today (31.06 above the re-based 28.00 cap) and the signal flipped from the breakdown family to ReversalUp_Oversold, HOLD conviction 62. But HK-12 records that this exact pattern already failed once in this book - the 07-17 market-sell of 900 @44.80 was withdrawn on a ReversalUp_Oversold cv71 flip and 7709 went to 42.52 anyway - so the flip is explicitly NOT treated as a reason to cancel. Ledgered optional so the counterfactual grader prices what not cutting costs. Position -72.9% from a 114.56 cost, 1.3% of book value. Stop 28.00 stands; the model's 7.75 is HK-12 degenerate (-75%) on a 37.5%-of-price ATR. |
| 0316 | OOIL | 148.60 | WATCH | 146.26 | 140.94 | 150.11 (+2.6%) | 151.54 (+3.6%) | 170.37 (+16.5%) | The strongest technical buy case in the book and still not actionable. OOIL prints BUY at conviction 73 on the book's single most reliable buy pattern (PullbackBuy_VWAP, +5 scorecard bias, hit rate.75 over 10 graded trades), ranks first on the buy side, and screens UNDERVALUED - adjusted conviction 81, the highest reading of the day. It is blocked by the quality gate, and the gate is right: the shipping line is burning cash (free cash flow -HK), revenue is shrinking 9.2% year on year, and its quality rank is 21.. A cheap, deteriorating cyclical is not the same trade as a cheap, compounding one. OPERATIONAL: Quality_Pass False is a hard block on NEW entries only. Would also have queued rather than bought - Entry_Distance_ATR -0.50 puts the close 0.5 ATR above the 146.26 model entry. Re-arm: a Quality_Pass flip on the next fairvalue refresh, with the PullbackBuy print intact. · news |
| 0941 | CHINA MOBILE | 81.85 | WATCH | 81.85 | 81.52 | 81.81 (-0.0%) | 81.77 (-0.1%) | 80.36 (-1.8%) | Screens cheap but the trade does not clear the bar twice over. The signal family is the book's chase-grade pattern (FailedBreakdown_Up_20D, -4 bias, hit rate.30 over 12 graded buys), taking adjusted conviction to 62 against a 65 floor, and the quality screen fails outright (F-Score 3, quality rank 14) which independently blocks a new entry. The model's own forward view is flat to negative anyway - 4-week 81.77 and 12-week 80.36 both sit at or below the 81.85 close - so there is no return case to override either objection. |
| 0005 | HSBC HDGS(USD0.50) HK O'SEAS | 161.20 | HOLD | 161.20 | 159.50 | 160.63 (-0.4%) | 159.71 (-0.9%) | 163.31 (+1.3%) | Retired to a WATCH line on 08-06 after three unexecuted sell cards (HK-11, escalation spent) and the retirement holds: the close recovered to 161.20, back above the 159.50 line, and the label is a bullish HOLD. The bank is armed with one profit-take tick firing (12-week projection 163.31 below 1.02x the close), but re-issuing a sell the session after retiring it is exactly what HK-11 forbids, and the arm rests only on T12_Progress, which is >=90 on every held row in this book and carries no information. Trail-raise only. Ex-dividend 13-Aug will gap the shares ~0.5% mechanically - that is NOT a stop break. Re-arm: any settled close below 159.50. |
| 0700 | TENCENT | 478.80 | HOLD | 478.80 | 466.94 | 473.99 (-1.0%) | 475.76 (-0.6%) | 499.99 (+4.4%) | Armed with a conviction-collapse tick firing as the engine dropped to NO TRADE on consensus conflict), but Q2 results land 12-Aug - three trading sessions out, inside the catalyst-aware-hold window - so the rule is hold and trail through the print, never pre-trim into it. The 12-week projection 499.99 is still 4.4% above the 478.80 close, so the forward case is intact. Stop 466.94 stands, tighter than the standing overlay's ~430 line. Close remains below EMA200 503.84; a break of 466.94 after the print is the exit trigger. |
| 2513 | Z.AI (KNOWLEDGE ATLAS) | 1246.00 | HOLD | 1246.00 | 862.00 | 1222.91 (-1.9%) | 1167.49 (-6.3%) | 1207.32 (-3.1%) | Retired to a WATCH line yesterday as the final HK-11 issuance, and today the tape and the model both back standing down: the shares jumped 14.6% to 1246 and the label decayed from STRONG SELL - REDUCE at conviction 73 to AVOID at 63 with portion 0 and no sell rank. Not selling yesterday's card at 1087 was worth HKon this line alone. Two things now argue hold rather than cut: the Hang Seng quarterly index review announcement is 14-21 Aug, five sessions out and inside the catalyst window with Z.AI a named inclusion contender, and the one profit-take tick that fires (12-week 1207.32 below the close) is the known degenerate-projection artifact. Against that, a fresh US$4bn share sale is a real dilution overhang and the stock trades ~49% above consensus target - reasons to stay at one lot and not add. Re-arm at a settled close below 862.00; never re-issue on signal alone. |
| 0522 | ASMPT | 161.60 | HOLD | 161.60 | 145.50 | 165.03 (+2.1%) | 180.69 (+11.8%) | 181.63 (+12.4%) | Still the strongest forward view in the book - 4-week 180.69 is +11.8% and 12-week 181.63 is +12.4% over the 161.60 close - against a NO TRADE label at conviction 0 for a second straight session. Repeated NO TRADE is a profit-take tick, but the position is 17.7% underwater from a 196.35 cost and 3.0 ATR below it, so nothing is armed on gain and trimming here would be selling the recovery. Hold 1,100, stop 145.50. TSMC July revenue 10-13 Aug and AMAT FQ3 13-Aug read through to this name. |
| 1347 | Hua Hong Semiconductor | 141.20 | HOLD | 141.20 | 118.00 | 138.99 (-1.6%) | 151.25 (+7.1%) | 148.32 (+5.0%) | Declined on merit, not displaced. The only profit-take tick that fires is repeated NO TRADE; the 12-week projection 148.32 sits above 1.02x the close, and the 4-week view is +7.1%, so the forward case is positive and the shares are well above their 200-day average (141.20 vs 114.36) after a +3.3% session. Q2 results are now behind it. Trimming a name into a rising forward view on a label-absence tick alone is the error this book made with the 07-15 re-entry. Hold 3,000, stop 118.00. Position -11.8% from a 160.00 cost. |
| 2888 | STANCHART | 232.20 | HOLD | 232.20 | 224.83 | 234.08 (+0.8%) | 231.91 (-0.1%) | 224.39 (-3.4%) | Armed with two ticks firing - dead 12-week projection (224.39 below the 232.20 close) and repeated NO TRADE - so a 50% trim is technically due, but at it banks only ~HKand loses the third dispose slot to CATL on size. Ledgered so the counterfactual grader prices the miss. The shares sit well above their 200-day average (232.20 vs 190.28) and RSI 59.8 is not extended. Hold 348, stop 224.83, which is 1.1x ATR under the close. |
| 0823 | LINK REIT | 38.60 | HOLD | 38.60 | 37.84 | 38.70 (+0.3%) | 38.71 (+0.3%) | 39.32 (+1.9%) | Two ticks nominally fire but both are noise on this name: the dead-12-week test clears by HKon a REIT whose ATR is 1.5% of price, and the second tick is simply the absence of a label. The forward view is flat, not negative (1-week +0.3%, 4-week +0.3%, 12-week +1.9%), and the shares hold above their 200-day average. Not a trim case. Hold 2,800, stop 37.84. |
Outcome & track record
Accuracy at the time of this record.
148 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 →
-0.249 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.