Recommendations
0 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 |
|---|---|---|---|---|---|---|---|---|---|
| 3988 | Bank of China | 5.95 | SELL 50% | 5.95 | 5.79 | 6.00 (+0.9%) | 5.96 (+0.3%) | 5.93 (-0.3%) | Bank of China is the book's biggest winner at +9.3% and its conviction score collapsed forty points today as the stock gave back most of Friday's rally. We banked 30% at this morning's open at 6.10 and now take half of what remains, holding the rest behind an unchanged stop - a second profit-take into deterioration, not an exit, since the signal is not bearish and the model still shows no downside over twelve weeks. ADR-0012 armed on the GENUINE leg (Gain_ATR 3.22 >= 3.0 vs a real Avg_Cost 5.44), so HF-2 does not reach this card. TWO ticks: (b) conviction, -40 pts, label MomentumContinuation_Up -> FailedBreakdown_Up_20D; (c) Target_12Week 5.93 < 1.02 x 5.945 = 6.0639. (a) cleared - RSI_DIV gone overnight, PROFIT_ARMED standalone. 2 ticks -> profit_take_2 = 50%. G1 inapplicable (no longer a fresh Breakout/Momentum at cv>=80) and (b)/(c) fire regardless. SENSITIVITY DISCLOSED: tick (c) fired on holdings again tonight and is non-selecting on this book; strip it and this is 1 tick -> 30%. Valuation NA_CCY -> val_adj 0, no tier bump. Bias 0 (FailedBreakdown_Up_20D|sell n=5 actionable, hit_1w 0.50). Action price = anchor close, provenance band 5.911-5.989 (Entry 5.95 +/- 0.25 x ATR 0.157). at board lot 1,000 = HKrealised +HK; residual 7,000. STOP 5.79 UNCHANGED = max(live 5.79, engine 5.64, chandelier 5.631) - engine and chandelier both sit below the standing stop after today's fall; cushion 2.61% / 0.99 ATR. L8: executes at the 2026-09-08 open, reprice if it gaps >~1.5%. · news |
| 0939 | CCB - H SHS | 9.53 | SELL 50% | 9.54 | 9.37 | 9.56 (+0.3%) | 9.72 (+1.9%) | 9.68 (+1.5%) | China Construction Bank has reached 98.5% of the model's twelve-week objective while its conviction score fell forty points in one session, and it is the name in this book most exposed to Wednesday's China inflation print. We take half off at a +3.6% gain and carry the balance into the data. The arm is GENUINE, not the 2888/3968 artifact: T12_Progress 98.50 with a NON-inverted Est_12W (9.68 vs 9.535 close, +1.52%), so HF-2's inversion suppression is inapplicable; Gain_ATR 1.24 is below the 3.0 bar, which is why the T12 leg had to be checked. TWO ticks: (b) conviction, -40, MomentumContinuation_Up -> FailedBreakdown_Up_20D, tracking a real -2.30% fall; (c) 9.68 < 1.02 x 9.535 = 9.7257, by only 0.47%. 2 ticks -> profit_take_2 = 50%. STATED PLAINLY: the forward case is thinner than 3988's - the 12wk sits +1.52% ABOVE the close and the 4wk +1.94%, so the model still sees upside; what justifies the card is a genuine near-target arm plus a 40-point conviction break, not a dead forward view, and the optional_trim [saved] n=85 counterfactual is the standing argument against it. Valuation NA_CCY -> 0. Bias 0. Action price = anchor close, provenance band 9.463-9.597 (Entry 9.53 +/- 0.25 x ATR 0.2694). at board lot 1,000 = HKrealised +HKvs cost 9.20; residual 3,000. STOP 9.37 UNCHANGED = max(live 9.37, engine 8.99, chandelier 8.996); cushion 1.73% / 0.61 ATR, the TIGHTEST in the book - itself part of the case for halving ahead of a high-confidence macro print. EVENT LOAD, heaviest in the book: 0939 is named on four Event Horizon rows inside 10 sessions - China Aug CPI/PPI 09-09 (high conf, +/-1.5%, PPI governs the China bank multiple), China TSF ~09-10, China activity data 09-15, LPR fixing 09-21 (stance queue-below; a 5Y cut compresses lending margin). L8: executes at the 2026-09-08 open. · news |
| 3328 | Bank of Communications | 7.88 | SELL 30% | 7.88 | 7.61 | 7.86 (-0.3%) | 7.96 (+1.0%) | 7.91 (+0.4%) | Bank of Communications is 99.6% of the way to its twelve-week objective at a +5.1% gain, so we take a further ~29% off and hold the rest. This is the smallest and least-supported card in the verdict and is ranked last for that reason. The arm is GENUINE - T12_Progress 99.62 with a NON-inverted Est_12W (7.91 vs 7.88, +0.38%), HF-2 inapplicable. ONE tick, and it is the weak one: (c) Target_12Week 7.91 < 1.02 x 7.88 = 8.0376. (a) does NOT fire - RSI_DIV CLEARED overnight (it drove 09-04's 50% tier), row now prints PROFIT_ARMED alone; (b) does not fire - conviction is -14, below the 20-pt bar, the only holding whose conviction did not break; (d) does not fire. 1 tick -> profit_take_1 = 30%, the minimum tier. STATED PLAINLY: this card rests entirely on tick (c), which fired again tonight and which this book's systemic findings have called non-selecting for four sessions; RSI has cooled 72.7 -> 65.6. If tick (c) were correctly specified for low-dispersion names this card would not fire at all - it ships because the rule is the rule, at the minimum tier, on the book's smallest position. Valuation NA_CCY -> 0, no tier bump. Bias 0. Action price 7.88 = anchor close = Entry_Price exactly. Volume: 30% x 3,500 = 1,050 -> board lot 1,000, FLOORED not rounded up -> = 28.6% effective = HKrealised +HKvs cost 7.50; residual 2,500. STOP 7.61 UNCHANGED = max(live 7.61, engine 7.51, chandelier 7.512); cushion 3.43% / 1.47 ATR, the WIDEST in the sleeve - another reason for the minimum tier. L8: executes at the 2026-09-08 open. · news |
| 2888 | STANCHART | 239.60 | HOLD | 239.60 | 232.00 | 237.91 (-0.7%) | 233.79 (-2.4%) | 232.66 (-2.9%) | Standard Chartered was the only name in the book to gain today and the only one whose conviction rose, printing a maximum-conviction score at a 52-week-high area - so we hold the full line and raise the protective stop instead of trimming, tightening risk ahead of a genuinely undecided Fed meeting. HF-2 FALSE ARM, FIFTH CONSECUTIVE SESSION: T12_Progress 102.98 is produced by an Est_12W (232.66) sitting 2.90% BELOW the 239.60 close, not by a winner near objective - the position is +4.80% with Gain_ATR 2.32, below the 3.0 bar. Tick (a) RSI_DIV is present and G1's family list does not literally cover AnticipationUp_VCP, so the tick would fire at full force; HF-2 suppresses the ARM itself, and HF-2's required independent bearish basis is absent (no formal SELL; 1.60 ATR of stop cushion). Trimming a cv100 name making highs on a divergence is the NBIS error. HONEST CAVEAT, unchanged from 09-04: HF-2's text conditions on oversold/bounce-family signals and this is neither - the MECHANISM is being applied past the letter of the family clause for a fifth session (systemic finding 6). STOP RAISED 229.89 -> 232.00 = max(live 229.89, ENGINE 232.00, chandelier 230.126); cushion 3.17% / 1.60 ATR. This executes the Event Horizon's explicit tighten-trail stance on 2888 into the 17 Sep FOMC (+/-4.5%, 50.6% hold vs 49.4% for a 25bp HIKE, with the HKMA following mechanically the same HK morning at +/-2.0%) - and note the 16 Sep close is the last HK session before the Fed prints overnight, so every protective level must be standing by then. Stale overlay's '0005/2888 hold unless below the recorded stop' is satisfied. Position is HK = 52% of the equity book. · news |
| 3968 | China Merchants Bank | 51.90 | NO ACTION | 51.90 | 50.88 | 51.66 (-0.5%) | 51.78 (-0.2%) | 50.91 (-1.9%) | China Merchants Bank fell 2.7% today and its conviction score dropped forty points, but the profit-take that would normally follow is suppressed because the position is not genuinely near its objective - the reading that makes it look so is an artifact of a twelve-week estimate sitting below the current price. We hold the balance behind an unchanged stop; the line was already cut this morning. THE MOST DEFEASIBLE CALL IN THE VERDICT, flagged as such: T12_Progress 101.94 is an INVERSION ARTIFACT (Est_12W 50.91 = -1.91% vs the 51.90 close) with Gain_ATR 2.40 below the arm bar -> HF-2 suppression. BUT tick (b) genuinely fires (cv, the SAME -40 that trims 3988 and 0939) and tick (c) fires too - so this position is held on HF-2's LETTER against two live deterioration ticks, because a conviction collapse is neither of the two independent bearish bases HF-2 names (decisive close below stop; fresh high-conviction formal SELL). Mitigating: the line was already cut at this morning's open @53.25, 2.60% above tonight's close, so it has been de-risked once today. If HF-2 gains a 'conviction drop >= 20 pts' clause at the next learnings review this becomes a trim - that is systemic finding 3, the highest-value new item in the run. Ledgered optional:true so the counterfactual grader prices what not trimming cost or saved. STOP 50.88 UNCHANGED = max(live 50.88, engine 49.19, chandelier 49.186); cushion 1.97% / 0.75 ATR. News: none found - Yahoo returns profile/quote pages only, no September event. · news |
| 2388 | BOC Hong Kong Holdings | 52.80 | BUY-STOP | 53.21 | 49.96 | 52.61 (-1.1%) | 53.24 (+0.1%) | 53.50 (+0.5%) | BOC Hong Kong is the one name in the book whose twelve-week outlook is genuinely above today's price, but the model's one-week view sits below the level we would have to pay to get in, and the stock fails our quality screen - so we stand aside rather than pay up. Print gate PASSED (ANTICIPATE 09-04 + 09-07 = 2 of last 3) and cv 68 >= the 55 floor. It owns THE ONLY non-inverted 12-week estimate on any bullish row in the book (Est_12W 53.50 vs 52.80 close, +1.33%) - the single name HF-3 clears. Blocked on two other screens: HF-4 FIRES (Est_1W 52.61 < Entry_Price 53.21, -1.13% - the estimation layer contradicting the conviction layer, the signature that marked the 2888 07-17 bad fill), and the QUALITY GATE is LIVE and FAILING (OVERVALUED_VS_RI is a non-NA_* row with Quality_Pass False) -> hard block on a new entry. val_adj -2 (MoS_FV -0.3135 x 5). Conviction also decayed. Moot regardless: one anticipate slot and 0005 outranks it. NOTE THE SYMMETRY - 0005 clears HF-4 and fails HF-3; 2388 clears HF-3 and fails HF-4. Neither is available and no combination of them is. Re-arm: Est_1W >= Entry_Price AND Quality_Pass True. |
| 0966 | China Taiping Insurance | 21.86 | WATCH | 20.84 | 19.80 | 21.81 (+4.7%) | 21.43 (+2.8%) | 20.67 (-0.8%) | China Taiping screens as the cheapest name available and tops the buy ranking, but it failed our balance-sheet quality screen and reversed hard today, closing 3.5% down after touching a fresh high - so we stand aside for a fifth session rather than catch it. Adjusted conviction 65 = 60 raw + val_adj +5 (MoS_FV +0.786 -> +10, HALVED for the quality failure) + bias 0, i.e. EXACTLY at the floor - so as on 09-03 and 09-04 it is not the floor that stops it. QUALITY GATE, hard: Quality_Pass False (F-Score 5) on an UNDERVALUED - i.e. NON-NA_* - row means the gate is LIVE and blocks a new entry outright. HF-3 also rejects: Est_12W 20.67 vs the 21.86 close = -5.44% inverted, the deepest inversion in the scan. HF-4 is the one screen it passes (Est_1W 21.81 > Entry 20.84). And today's bar IS the falling-knife shape HF-3 was written for: open 23.30, high 23.36, low 21.76, close 21.86 - a -3.53% outside-down reversal off a fresh high, with the model's response being to drop the entry to a 20.84 limit, 4.7% BELOW the close. L1's 'favor PullbackBuy_VWAP' is a ranking preference, not a licence to override a hard gate. Contrast with 0005's -0.25% inversion the same evening: a materiality band on HF-3 would separate these two correctly and block only this one (systemic finding 4). Re-arm: Quality_Pass True, or a non-inverted Est_12W. |
Outcome & track record
Accuracy at the time of this record.
31 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 →
5.000 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.