Recommendations
0 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 |
|---|---|---|---|---|---|---|---|---|---|
| 1347 | Hua Hong Grace Semiconductor | 139.90 | SELL 30% | 139.90 | 122.67 | 141.55 (+1.2%) | 151.11 (+8.0%) | 148.40 (+6.1%) | Hua Hong bounced 7.4% with the whole chip sector today, which takes the stop-loss case off the table - but the position is still down 12.6%, still below its 20- and 50-day averages, and the model now sees only about 6% of upside left to its 12-week target. We bank a third of it at 139.90 and keep the other for the semiconductor cycle, with a protective stop at 122.67. OPERATIONAL: ADR-0012 profit-take, ARMED (T12_Progress 94.27) + tick (d) fires on SIX consecutive NO TRADE prints (10:30/11:30/14:00/15:00/15:30/17:00). One tick -> profit_take_1 = 30%, minimum tier. Adjusted conviction 1 = 0 (NO TRADE) + 0 (valuation) + 1 (NoTrade|sell, n=10, ACTIONABLE); a fired profit-take is its own origination class and is not gated by the 65 floor. NoTrade_Reason verbatim: 'Consensus conflict + low conviction'. The 08-14 STOP-BREAK thread is RETIRED today (close 139.90 back above the 131.49 trail, 12wk positive) - this card replaces it on a DIFFERENT mechanism, it is not an HK-11 re-issue. Valuation amplify (MoS -17.569 would bump SUPPRESSED by G2: DCF_per_share blank, Quality_Pass False. KNOWN DEFECT, stated: PROFIT_ARMED on a -12.56% position (Gain_ATR -1.64) is the false-arm systemic finding, fifth session running; fired anyway at minimum tier per HK-5 (optional_trim n=53, wh_4w +3.70%, 'saved') and the ADR-0012 ban on prose-only advice. Board lot 100. Remaining 2,100 trail 122.67 (2.0xATR chandelier off the 147.20 swing high, 1.40xATR below close, inside the HK-12 sanity band). Catalysts: SK 1-20 Aug semi exports 21-Aug, NVIDIA Q2 26-Aug - both argue for keeping the majority. · news |
| 0823 | Link REIT | 38.80 | SELL 30% | 38.80 | 37.97 | 38.86 (+0.2%) | 38.82 (+0.1%) | 39.86 (+2.7%) | Link REIT has gone nowhere for five weeks - it moved 0.05% today and the model's own four-week estimate is 38.82, which is also 0.05% away. That is dead money tying up capital, so we take 30% off at 38.80 ahead of the Fed minutes and Hong Kong CPI, both of which hit a rate-sensitive REIT hardest. The remaining 2,000 units keep a stop at 37.97. OPERATIONAL: ADR-0012 profit-take, ARMED (T12_Progress 97.34) + tick (d) fires on an unbroken NO TRADE run through every print in the 7-snapshot window back to 08-13. One tick -> profit_take_1 = 30%. Adjusted conviction 1 = 0 + 0 (MoS -0.024, no bump) + 1 (NoTrade|sell). NoTrade_Reason verbatim: 'Consensus conflict + low conviction'. HK-11 THREAD POSITION #2 - this is the 08-14 card re-issued once, which the rule permits; if unexecuted again, 08-18 is #3 = escalate-then-retire and this trim goes to a WATCH line. Name is quiet not broken: close 38.80 at/above EMA20 38.78 / EMA50 38.55 / EMA100 38.36 / EMA200 38.21, RSI 51.37, ATR 1.5% of price. 12wk 39.86 EXCEEDS the DCF anchor 37.84 on a GROWTH_PRICED_IN name - projection assumes growth beyond what is priced in. Sizing 30% x 2,800 = (board lot 100). Trail raised 37.86 -> 37.97 (2.0xATR off the 39.10 high). Catalysts: FOMC minutes 19-Aug (HK reaction 20-Aug), HK July CPI 21-Aug. · news |
| 0700 | Tencent | 446.40 | HOLD - exit retired to WATCH 430 | 446.40 | 430.00 | 446.42 (+0.0%) | 439.25 (-1.6%) | 439.87 (-1.5%) | Tencent is still below every moving average and still below our stop, but we have now written the same exit three sessions running without it being executed, on a position worth about 1% of the book. Rather than write it a fourth time we set a hard line: if it closes below 430 we are out. Kuaishou's results on Wednesday and Alibaba's on Thursday are the read-throughs that will settle it. OPERATIONAL: HK-11 POSITION #3 - thread ran 08-13 (#1 MOO) -> 08-14 (#2 MOO, the single permitted escalation, spent) -> today RETIRED to a WATCH line, exactly as the 08-14 card pre-committed in writing. THE EXIT CASE IS STILL INTACT: close 446.40 is -4.4% below the 466.94 stop (fourth consecutive close below), 12wk 439.87 = -1.5% NEGATIVE, so the L2 combo is complete; below EMA20 462.09 / EMA50 462.12 / EMA100 476.74 / EMA200 501.39, RSI 43.42. This is a BOOKKEEPING retirement, not a signal decay, and is recorded as such. Justification: = HKthe smallest holding in the book, and HK-11 was quantified at a -HKexecution gap from exactly this re-writing behaviour. RE-ARM: decisive close < 430.00 (the news overlay's own standing 0700 override, verbatim) OR a fresh SELL-REDUCE at a settled close. Recorded stop moves 466.94 -> 430.00 so the monitor tracks the live line. Avg_Cost 1.00 is a data artifact (+44,540%, Gain_ATR 34.06) - PROFIT_ARMED on this row is meaningless and was ignored. Inaction : the 08-14 card would have sold 200 @444.80 (open); at 446.40 that non-execution cost -HK. · news |
| 0316 | OOIL | 158.90 | NO ACTION | 157.82 | 142.58 | 159.96 (+1.4%) | 164.61 (+4.3%) | 171.60 (+8.7%) | OOIL is the strongest buy signal this book has produced in weeks - a breakout to a 52-week high on the back of container freight rates hitting their 2026 peak, with Shanghai-US East Coast spot at a box, up 19% in three weeks, and Maersk raising guidance twice in six weeks. We are still not buying it: the company's trailing accounts show negative free cash flow and a 9% revenue decline, which fails our quality screen outright, and the stock closed 7.3% up on its own high, which is the kind of chase this book has repeatedly lost money on. We would rather wait for a pullback. OPERATIONAL: BLOCKED BY THE HARD QUALITY GATE - Quality_Pass False on F_Score 5, FCF -US (FCF_Yield -0.02%), Rev_Growth -9.2%, Earn_Yield 1.6%, Quality_Rank 20.5/31. Gate blocks NEW ENTRIES only; 0316 is flat so it is blocked. Adjusted conviction 91 = 90 + 3 (UNDERVALUED MoS +0.299, val_adj HALVED because Quality_Pass is False) - 2 (BreakoutUp_20D|buy, n=10, hit_1w.40, hit_4w.125, ACTIONABLE). BLOCKED INDEPENDENTLY BY HK-1: BreakoutUp_20D is one of the three HK momentum-chase families ruled non-actionable without confirmation; close 158.90 after +7.29%, the 160.00 intraday high IS the 52W high, RSI 69.60 - HK-1 demands RSI < ~70 plus a pullback to the model entry or a second print. TENSION STATED: the gate reads TTM fundamentals dated 2026-08-11, a window that PREDATES the freight inflection which is the entire investment case; the trailing revenue decline is what a cyclical looks like at the bottom. Applied anyway - hard gate, no mid-run policy change. Logged as systemic finding #1. WATCH LINE (both legs required): a settled-close BUY with RSI < 65 after a pullback to ~157.82, AND Quality_Pass turning True on a refreshed fairvalue run that captures the freight cycle. Ledgered optional:true so the counterfactual grader prices what declining it cost. · news |
| 2513 | Z.AI (Knowledge Atlas) | 1203.00 | HOLD - trail raised | 1203.00 | 1126.38 | 1212.80 (+0.8%) | 1150.96 (-4.3%) | 1265.19 (+5.2%) | Z.AI gave back 5.3% but MSCI has confirmed it as one of the three largest additions to its Emerging Markets index in the August review, effective at the close on 31 August - that means mechanical index buying is coming, and the Hang Seng review lands on Friday with maiden results due by month-end too. Selling into that stack of catalysts is how you give away the upside you were positioned for. We hold and raise the stop to 1,126. OPERATIONAL: ARMED (T12_Progress 95.08) but NO tick fires - (a) no second warning code, (b) conviction fell only (-12, under the 20-pt bar), (c) does not fire (12wk 1,265.19 > 1.02 x 1,203 = 1,227.06), (d) label AVOID not NO TRADE. Printed STRONG SELL - REDUCE at 10:30/11:30/14:00 (conv 37/39/39) and DECAYED to AVOID 42 into the settled close - L4 makes the close authoritative, so no sell card. MANDATORY trail-raise 1,114.23 -> 1,126.38 (2.0xATR chandelier off the 1,435 high). CATALYST HOLD, three inside the window: Hang Seng quarterly review announcement 21-Aug, maiden interims due on/before 31-Aug under LR 13.49(6), MSCI effective 31-Aug. Trim only on a confirmed lower-high or a decisive stop break. HK-12: model stop 1,511.62 is 26% ABOVE the close - degenerate, rejected (fifth consecutive session). · news |
| 0005 | HSBC Holdings | 163.50 | HOLD | 163.50 | 159.50 | 163.84 (+0.2%) | 161.78 (-1.1%) | 165.54 (+1.2%) | HSBC recovered 1.7% and is now the strongest hold in the book on conviction, within 4% of its 52-week high. Our exit line here is a close below 159.50 and it has not been hit, so we hold all. OPERATIONAL: ARMED (T12_Progress 98.77) + tick (c) fires (12wk 165.54 < 1.02 x 163.50 = 166.77). A 30% trim worth ~HK - the LARGEST candidate in the book - again NOT taken: RETIRED 08-06, re-arm 'close below 159.50' NOT fired at 163.50, HK-11 binds. Conviction RECOVERED. Stop held at the 159.50 watch line (tighter than the 158.88 chandelier). FOMC minutes 19-Aug, US flash PMIs 21-Aug, HK July external trade 25-Aug. |
| 0522 | ASMPT | 182.10 | HOLD | 182.10 | 163.69 | 185.78 (+2.0%) | 198.84 (+9.2%) | 203.46 (+11.7%) | ASMPT rose 3.5% and still carries the best forward view in the book - roughly 9% to the four-week estimate and 12% to the twelve-week. The signal's conviction dropped sharply, which we are watching, but there is no profit to take here and no exit trigger, so we hold and keep the stop at 163.69. OPERATIONAL: the ONE held name with NO arm at all (T12_Progress 89.50 < 90, Gain_ATR -1.44). Conviction FELL (-23), which WOULD be tick (b) - but with no arm there is no profit-take to fire. Flagged as the sharpest deterioration in the book. Close 182.10 well above EMA20 168.26 / EMA50 171.11. Trail HELD 163.69 (tighter than the 163.17 chandelier off the 182.90 high); model stop 161.51 is looser, trail wins. NVIDIA 26-Aug reads through. |
| 2888 | Standard Chartered | 238.00 | HOLD - trail raised | 238.00 | 229.05 | 238.29 (+0.1%) | 235.85 (-0.9%) | 230.44 (-3.2%) | Standard Chartered closed at 238, a whisker off its 52-week high, and yesterday's worrying conviction collapse reversed completely in one session. We hold and raise the stop to 229.05, which sits just under our 230.19 exit line - one decisive down session resolves this either way. OPERATIONAL: ARMED (T12_Progress 103.28, above its own 12wk target) + tick (c) fires (230.44 < 1.02 x 238.00 = 242.76). NOT taken: RETIRED 08-13, re-arm 'decisive close < 230.19 OR fresh SELL-REDUCE' NOT fired at 238.00 with no SELL-REDUCE print. HK-11 binds. Yesterday's flagged 'sharpest conviction collapse in the book' REVERSED:. Trail raised 228.47 -> 229.05 (2.0xATR off the 239.20 high). FOMC minutes 19-Aug, US flash PMIs 21-Aug. |
| 0981 | SMIC | 75.15 | HOLD - trail raised | 75.15 | 68.53 | 75.63 (+0.6%) | 78.21 (+4.1%) | 79.01 (+5.1%) | SMIC added another 6.1% as Chinese foundries extended the global chip rebound. Retiring the sell thread here on Friday keeps paying - the trade we did not make is now worth about HKto us. We hold and raise the stop to 68.53. OPERATIONAL: ARMED (T12_Progress 95.11) but NO tick fires - (c) does not fire (12wk 79.01 > 1.02 x 75.15 = 76.65), (b) conviction ROSE, (d) label AVOID not NO TRADE. RETIRED to a WATCH line 08-14 at 65.16; close 75.15 is nowhere near it. Trail raised 65.16 -> 68.53 (2.0xATR off the 76.60 high). Inaction credit: the 08-13 card would have sold 2,000 @67.55; at 75.15 that is +HKbetter unsold. SK 1-20 Aug semi exports 21-Aug, NVIDIA 26-Aug. · news |
| 7709 | XL2CSOPHYNIX (CSOP SK Hynix 2x) | 40.80 | HOLD | 40.80 | 28.00 | 41.75 (+2.3%) | 39.66 (-2.8%) | 41.76 (+2.4%) | The leveraged SK Hynix tracker jumped another 10.4% as memory names extended their rebound - Samsung and SK Hynix were up 6.2% and 4.1%. Retiring the sell thread on Friday is being vindicated a second session running. We hold, with the exit line unchanged at a close below 28.00. OPERATIONAL: ARMED (T12_Progress 97.70) but NO tick fires - (c) misses by 0.14 (12wk 41.76 > 1.02 x 40.80 = 41.62), (b) conviction = -16, under the bar, (d) label AVOID not NO TRADE. Printed STRONG SELL - REDUCE 70 at 15:30 and DECAYED to AVOID 57 at the settled close - L4 governs, so the 'fresh SELL-REDUCE' leg of the re-arm is NOT triggered. RETIRED 08-14 to the 28.00 WATCH line; close 40.80. HK-12: model stop 53.05 is 30% ABOVE the close - degenerate, rejected; ATR 6.12 is 15% of price so no ATR-based stop can bind on this product. WATCH line HELD at 28.00. Leveraged-2x rule (feedback_leveraged_etf_exits) independently caps any future action at a 50% trim, never a full exit on a reversal-family signal. Inaction credit: the 08-13 card would have sold 900 @33.58; at 40.80 that is +HKbetter unsold. US monthly options expiry 21-Aug, SK exports 21-Aug. · news |
| 0939 | CCB - H Shares | 8.90 | HOLD - trail raised | 8.90 | 8.61 | 8.86 (-0.4%) | 8.95 (+0.6%) | 8.87 (-0.3%) | CCB firmed 1.8% with conviction improving, and our exit line of a close below 8.46 is nowhere near. We hold and nudge the stop to 8.61. The loan prime rate fixing on Thursday and interim results on 28 August are the next things that move it. OPERATIONAL: ARMED (T12_Progress 100.34) + tick (c) fires (8.87 < 1.02 x 8.90 = 9.078). RETIRED 08-11; re-arm 'fresh SELL-REDUCE or decisive close < 8.46' NOT fired at 8.90, label HOLD with conviction. HK-11 binds. Trail raised 8.52 -> 8.61 (2.0xATR off the 8.955 high). China LPR fixing 1Y+5Y 20-Aug; interims + interim dividend 28-Aug; China Jan-Jul industrial profits 27-Aug. |
| 3750 | CATL | 652.00 | HOLD | 652.00 | 614.95 | 647.24 (-0.7%) | 641.75 (-1.6%) | 634.21 (-2.7%) | CATL added 1.6% with conviction improving. Our exit condition is a close below 625.50 and at 652 it has not been met, so we hold and keep the stop at 614.95. BYD's interims on 31 August are the sector read-through. OPERATIONAL: ARMED (T12_Progress 102.81) + tick (c) fires (634.21 < 1.02 x 652.00 = 665.04). RETIRED 08-11; re-arm 'close < 625.50 or fresh SELL-REDUCE' NOT fired, label HOLD is not a SELL-REDUCE. HK-11 binds. Trail HELD 614.95 (tighter than the 612.51 chandelier). China LPR 20-Aug, Jan-Jul industrial profits 27-Aug, BYD interims 31-Aug (first tradeable HK session). |
Outcome & track record
Accuracy at the time of this record.
172 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.