Recommendations
0 to acquire · 0 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 |
|---|---|---|---|---|---|---|---|---|---|
| 2888 | STANCHART | 236.00 | HOLD | — | 223.61 | 236.44 | 234.61 | 227.76 | Standard Chartered closed at HK, its best level since we bought it, and the position is now up 3.2% or about HKThe signal has recovered for a second straight session and is strengthening rather than fading, so we are not selling into it -- instead we are tightening the protective stop from HKto HK, which locks in more of the move while leaving the position room to keep working. The one thing we are watching is that the model's three-month projection still sits below today's price; that is a reason to keep the stop moving up, not a reason to sell today. OPERATIONAL: PROFIT_ARMED SUPPRESSED under ACTIVE HF-2 for the 11th consecutive session -- Target_12Week 227.76 < Close 236.00, inverted -3.49% and WIDENING for a 4th straight session (-2.62% -> -3.36% -> -3.49%); T12_Progress_Pct 103.62 is arithmetically 236.00/227.76 on an inverted target, and Gain_ATR is 1.19 against a 3.0 arm threshold. Fire-tick (c) reads fired (227.76 < 1.02 x 236.00 = 240.72) but fires off that identical inversion -- see systemic finding 1, tick (c) is logically implied by a T12_Progress-derived arm and cannot confirm it. HF-2's independent-bearish-basis test is NOT met: conviction RISING, HOLD label with no Sell_Rank, no non-PROFIT_ARMED Exit_Warning, close 2.00 ATR above the new stop, RSI 63.95. Ticks (a)/(b)/(d) all negative. The 08-07 post-execution trigger was checked leg by leg and none fired: (1) no close below 222.07 (236.00); (2) no other Exit_Warning code; (3) no fresh high-conviction SELL/REDUCE; (4) no fresh 2-consecutive-NO-TRADE run from 08-07 (HOLD, HOLD). TRAIL-RAISE is MANDATORY and fires the 08-07 pre-commitment exactly as written ('raise only when the 2.0-ATR chandelier exceeds the 222.07 ratchet floor, i.e. close > ~235.20'): chandelier = 236.00 - 2 x 6.194994 = 223.61 > 222.07, so stop 222.07 -> 223.61 (+1.54), ratchet only, never lowered. Peak stop re-bases to 223.61; ADR-0020 erosion 0%, no fire. The stop still sits HKBELOW the 228.63 cost, so a stop-out realises about -HK; residual stop-risk = (236.00 - 223.61) x 1100 = HK0.68% of book. Estimates ~80% coverage: 1wk 236.44 [224.13-252.43], 4wk 234.61 [205.61-269.85], 12wk 227.76 [193.24-296.50] -- a flat-to-down forward curve against a rising tape, which is the watch item, not a sell basis. NEXT TRAIL RAISE available when the chandelier clears 223.61, i.e. close > ~236.00 at current ATR. News: none found 08-07 -> 08-10; the HKdividend went ex 08-06 and is already absorbed under HF-5. · news |
| 3968 | China Merchants Bank | 49.00 | HOLD | — | 48.10 | 48.88 | 49.20 | 49.36 | China Merchants Bank is working: closed HK, up 0.7% on the day and now modestly ahead of our HKcost, with the signal recovering from last session's dip rather than deteriorating further. A profit-taking flag did appear on the position today, but on inspection it is an artefact of how the system measures progress -- it treats the shares as 99% of the way to their three-month target simply because that target sits less than 1% above the current price, when in reality this trade is one session old and up less than a percent. We are not trimming a barely-profitable week-old position on a measurement quirk. We hold, and the risk line stays at HK. OPERATIONAL: PROFIT_ARMED appeared for the FIRST time on this line and is SUPPRESSED -- and this is a NEW artefact class the ACTIVE playbook does not literally cover. The arm is NOT from gain (Gain_ATR 0.32 vs the 3.0 threshold); it is from T12_Progress_Pct 99.27, which the scan computes as Close/Target_12Week = 49.00/49.36. ADR-0012's stated criterion is '>=90% of the 12-week target' in the sense of progress OF THE TRADE, and true progress from cost to target is (49.00-48.65)/(49.36-48.65) = 49.3% -- barely half. The printed 99% would arm ANY position whose 12wk target sits within 11% of the close, irrespective of entry price or gain. Fire-tick (c) also cannot confirm this arm because it is mathematically implied by it: (c) is T12 < 1.02 x Close <=> Close/T12 > 98.04% => T12_Progress > 98 >= 90 => armed. Arm and tick are ONE input -- the exact degeneracy HF-2 names, but HF-2's literal trigger requires Target_12Week < Close and here Target_12Week 49.36 > Close 49.00 (+0.73%, NON-inverted), so HF-2 does not textually reach it, and HF-P15/HF-P1 (which would) are PROPOSED and INERT. Suppressed on the ADR-0012 arm criterion directly; escalated as systemic findings 1-3 and a learnings candidate for a third HF-2 widening covering the FLAT (non-inverted) case. Independent bearish basis: NONE -- conviction ROSE, label upgraded WATCH -> HOLD, no other Exit_Warning (tick a), conviction up not down (tick b), no NO TRADE print (tick d). HF-6 legs all clean. TRANSPARENCY on the 08-07 post-execution trigger: leg (2) as written said 'any Exit_Warning code appearing' and PROFIT_ARMED did appear -- it is NOT honoured as a trim trigger because leg (2) was authored in the HF-6 deterioration-code sense (RSI_DIV/CLIMAX_VOL/AVWAP_BREAK/EMA10_BREAK), whereas PROFIT_ARMED is the profit-take arm governed by the ADR-0012 fire rule evaluated above; legs (1)/(3)/(4) all negative. STOP UNCHANGED at 48.10: the anchor Stop_Price printed 46.81 -- the 2.0-ATR chandelier (49.00 - 2 x 1.0951) that PROFIT_ARMED itself triggered -- which is LOWER than the recorded stop, and the ratchet rule never lowers. 48.10 sits HKbelow the 48.65 cost so a stop-out realises about -HK; stop-risk = (49.00-48.10) x 2000 = HK0.09% of book. Next raise available when the chandelier clears 48.10, i.e. close > ~50.29 at current ATR. ADR-0020: stop 48.10 vs peak_stop 48.10 = 0% erosion, no fire. Estimates ~80% coverage: 1wk 48.88 [46.96-51.71], 4wk 49.20 [43.03-56.33], 12wk 49.36 [38.77-59.75]. Valuation NA_CCY -> val_adj 0, but the underlying screen stays the book's best: Quality_Pass TRUE, F-Score 5, Quality_Rank 4.5/11, MoS_FV +0.223. News: none found. · news |
| 0388 | HKEX | 416.80 | NO ACTION | 416.04 | 400.15 | 416.50 (+0.1%) | 416.06 (+0.0%) | 415.53 (-0.1%) | HKEX was the only name in the book generating a buy signal today and we are passing on it. The exchange screens as expensively priced against what its own business earns, it fails our balance-sheet quality test, and -- decisively -- the model's own three-month projection sits BELOW today's closing price. We would be paying up on a breakout for a stock the model expects to be worth less in three months, against a 4% downside risk line, one week before it reports interim results. The signal itself has also been unreliable, flickering on and off across the last four sessions rather than building. We pass. OPERATIONAL: rejected on THREE independent grounds, any one sufficient. (1) FLOOR -- conviction 64 + val_adj -6 + bias 0 = adjusted 58 vs the 65 floor, short by 7; the -6 is degree-scaled from MoS_FV -1.118 (OVERVALUED_VS_RI, rich against the exchange's own residual-income anchor). (2) HARD QUALITY GATE -- Quality_Pass FALSE (F-Score 4) bars a NEW entry outright; 0388 is unheld so the gate applies at full force. (3) HF-3 INVERTED 12WK -- Target_12Week and Est_12W are both 415.53 < close 416.80 (-0.30%): paying a 416.04 buy-stop for a modelled three-month destination below today's close, against a 400.15 stop (-4.0%), is exactly the payoff asymmetry HF-3 exists to refuse. Signal stability is poor: BUY 64 -> NO TRADE 0 -> NO TRADE 0 -> BUY 64, a flip-flop, not a build. HF-4 does NOT fire (Est_1W 416.50 > entry 416.04, +0.11%), so the 1-week view is the single healthy leg -- which is why this is a rejection rather than a red flag. NOT QUEUED: a GTC buy-stop cannot cure a hard quality-gate block, and RSI 67.96 into interim results on 08-19 would be buying an inverted forward view straight into an earnings print. RE-ARM TRIGGER (machine-checkable, HF-P12 form) -- revisit only when ALL FOUR hold on a settled close: Target_12Week > Close AND Est_12W > Close AND raw conviction >= 71 (the 65 floor plus the 6-point valuation haircut) AND Quality_Pass TRUE at the next fairvalue refresh. Scorecard cell BreakoutUp_20D|buy is n=2 (hit_1w 0.5, median_err_1w -0.99%, stop_first_rate 0.0) -- report-only at n<5, no bias applied, and note per HF-P5/HF-P17 that this book has never observed the parent-hk 'breakouts stop first' pattern. News: none found on Yahoo Finance / Seeking Alpha 08-07 -> 08-10. · news |
| 2388 | BOC Hong Kong Holdings | 50.55 | NO ACTION | — | 49.66 | 50.31 | 50.31 | 50.50 | BOC Hong Kong is the strongest unheld name in the book on conviction, but it is not generating an actual buy signal today -- it is on watch. It also still screens expensive against the bank's own book value, fails our quality test, and its three-month projection remains fractionally below the current price. The conditions we wrote down three sessions ago for changing our mind have not been met. We continue to pass, and we keep watching. OPERATIONAL: not actionable -- the label is WATCH, not a buy print, so 2388 never enters the buy pool regardless of its cv61 (the book's highest non-held conviction). Even if it did: adjusted 61 + val_adj -1 + bias 0 = 60 vs the 65 floor; Quality_Pass FALSE (F-Score 4, Quality_Rank 9.5/11) is a hard block on a new entry; and Target_12Week 50.50 < close 50.55 (-0.10%, inverted) trips HF-3. The 08-07 re-arm trigger -- 'both Target_12Week and Est_12W clear the close AND raw conviction >= 66' -- is UNMET on both legs (12wk inverted by 0.10%, conviction 61 vs 66). Third consecutive session withheld. Signal trail is choppy not building: WATCH 46 -> WATCH 50 -> BUY 56 -> WATCH 61. RE-ARM unchanged. Note the scorecard cell PullbackBuy_VWAP|buy is n=2 with ZERO graded outcomes -- L1 favours this family but this book has no evidence either way yet, so no bias is applied. News: none found. · news |
Outcome & track record
Accuracy at the time of this record.
6 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 →
0.184 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.
50 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.