Recommendations
1 to acquire · 1 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 |
|---|---|---|---|---|---|---|---|---|---|
| 0939 | CCB - H SHS | 9.14 | ADD — QUEUED @9.01 | 9.01 | 8.89 | 9.09 (+0.9%) | 9.21 (+2.2%) | 9.25 (+2.7%) | China Construction Bank flipped from NO TRADE to a BUY - ADD at conviction 71 on a PullbackBuy_VWAP setup - the one entry shape this book's own record ranks highest - with the buy limit sitting 0.74 ATR below the close, so it fills only on weakness and can never chase strength. The 12-week view (9.25) and the 1-week view (9.09) both sit above the entry, and the fair-value layer is blind here (mainland reporting currency), so the case is purely technical and it is a good one. The offsetting risk is timing: CCB publishes interim results tonight after the Hong Kong close with a two-sided plus/minus 4 percent reaction band on Monday, so a resting limit below the market fills precisely when the print disappoints. Operational nuance: adjusted conviction 71 = raw 71 + val_adj 0 + scorecard bias 0 (PullbackBuy_VWAP|buy is 1/1, report-only). FailedBreakdown_Up_20D|buy re-graded from bias -10 to +0 this run, lifting the de-facto buy moratorium. Quality_Pass reads False but does not apply - the gate blocks new entries only, never ADDs to held names. Volume = 43.88% x 6,000 held floored to CCB's 1,000-share board lot; cost HKagainst HKdry. HF-3 clears (Est_12W 9.25 > Close 9.14), HF-4 clears (Est_1W 9.09 > Entry 9.01), no-chase clears (Entry_Distance_ATR -0.74). The existing 6,000-share line keeps its peak stop at 8.89 - the engine's fresh 8.66 is the new tranche's own risk line and is NOT adopted for the line, which would be a stop reduction. Post-fill blended cost 9.1525. HARD L8 GAP-GUARD: if Monday's open gaps more than +/-1.5% (outside ~9.00-9.28) DO NOT place the order - re-adjudicate at the 31 Aug EOD; on a full -4% gap the fill would land below the 8.89 stop and be stopped out on arrival. The simultaneous PROFIT_ARMED + tick (c) on this row is suppressed: the position is at a loss (Gain_ATR -0.34) so the arm is the T12_Progress artifact leg, and the row itself prints BUY. Rationale for acting rather than waiting: the queue_expiry counterfactual is this book's one negative verdict (n=5, median +1.6% 1wk / +8.68% 4wk, verdict 'cost') - unfilled queues are what have actually cost this book money. · news |
| 3968 | China Merchants Bank | 49.02 | SELL 30% | 49.02 | 48.10 | 49.04 (+0.0%) | 49.28 (+0.5%) | 49.83 (+1.7%) | China Merchants Bank's signal deteriorated sharply - conviction fell 22 points to 50, the momentum pillar collapsed to 26, RSI dropped from, and the day printed a lower high, a lower low and the lowest close of the four-session window. That is a genuine loss of trend quality, not the measurement artifact that has suppressed every previous profit-take in this book, because the 12-week view here sits above the price rather than below it. Set against that, the shares must publish interim results within one trading session under Hong Kong's reporting deadline and this is the one holding with no sourceable results date and no event-calendar coverage - so the position is carrying an unpriceable event on a weakening signal. Cutting 30 percent reduces that exposure while leaving the majority of the line intact to participate if the print is good. Operational nuance: ADR-0012 profit-take, armed + tick (b) (conviction collapse >=20 pts). Tick (c) also fires (T12 49.83 < 1.02 x Close 50.00) but is deliberately NOT counted as a second tick - under the HF-2 mechanism the arm and tick (c) are one T12 relation read twice, consistent with the 08-21 / 08-25 / 08-26 rulings - so this is profit_take_1 (30%), not profit_take_2 (50%). HF-2 itself is silent because the arm is non-inverted (T12 49.83 > Close 49.02). G1 offers no cover (cv50, not a fresh cv>=80 breakout) and ticks (b)/(c)/(d) fire regardless of G1 in any case. The catalyst-aware escape hatch is met in substance, not just letter: confirmed lower high PLUS lower low PLUS lower close PLUS -22 conviction - compare 3988, where the same letter-level trigger fired on a rising close and was declined. Volume = 30% x 100; proceeds HKbanked gain HK; residual keeps the 48.10 stop (risk to stop HK). Valuation NA_CCY so no tier bump either way; scorecard cell FailedBreakdown_Up_20D|sell is n=3, report-only, bias +0. Close is still 1.2% above EMA200 (48.43) and Pillar_Consensus is still Consolidation_Bullish - which is exactly why this is a 30% trim and not an exit. Taken knowingly against the optional_trim counterfactual (n=58, median +1.72% saved at 1wk / +5.87% at 4wk) on the grounds that this is the first fire with a genuinely independent tick, and grading it is the only way that cell learns to separate artifact fires from real ones. Same +/-1.5% gap-guard applies: on an open outside ~48.29-49.76, reprice and re-confirm rather than executing blind. · news |
| 2888 | STANCHART | 230.40 | NO ACTION | 230.40 | 228.47 | 231.26 (+0.4%) | 231.97 (+0.7%) | 221.14 (-4.0%) | Standard Chartered closed up 0.35% and the position is back to +0.77%, but conviction is the lowest in the book at 36 and the 12-week projection sits 4% below the current price. The trim signal that fires on paper is an arithmetic artifact of that inverted projection rather than evidence of a failing position - the shares are still trending, still above their stop, and still labelled a hold - so the right response is to keep the risk line where it is and let the stop do the work. Operational nuance: ADR-0012 tick (c) fires (T12 221.14 < 1.02 x Close 235.01) but T12_Progress 104.19 is produced BY that same inversion, so arm and tick are one fact read twice - HF-2 requires an independent bearish basis (decisive close below stop, or a fresh high-conviction formal SELL) and neither exists. Tick (b) does not fire (-5 only). Stop cushion improved to HK = 0.38 ATR (was 0.204). Trail-raise yields nothing: 2.0x ATR chandelier is 220.27 off the close and 220.47 off the high, both far below the recorded 228.47, and stops never move down. PRE-ARMED: a decisive close below 228.47 is a FULL EXIT, not a trim - L2, stop break + Est_12W -4.02%. UK-domiciled so no results tonight, but fully exposed to the Jackson Hole reaction on Monday (+/-3.5% band). News none found - only the HKdividend that went ex on 6 Aug and undated chairman-succession chatter. · news |
| 3988 | Bank of China | 5.58 | NO ACTION | 5.58 | 5.44 | 5.57 (-0.2%) | 5.65 (+1.3%) | 5.74 (+2.9%) | Bank of China is the strongest line in the book - it closed up 0.45% at the highest close of the whole window, the position is at its best mark of +2.57%, conviction is unchanged and the 12-week view sits 2.9% above the price. A momentum-divergence flag technically triggers a 30% trim, but the shares publish interim results after tonight's close, so trimming now means selling into an unpriced two-sided event rather than reacting to one. The better trade is to hold the full line through the print and re-decide on Monday with the numbers in hand. Operational nuance: ADR-0012 tick (a) RSI_DIV fires - and today the divergence is finally observable for the first time (RSI 65.5 -> 64.9 into a higher close; the prior three sessions had RSI RISING into higher closes, which is why the flag was called unreproducible). The arm is legitimate and non-inverted (T12 5.74 > Close 5.58, Gain_ATR 1.34, the book's strongest). Declined on three grounds: (1) the 08-26 pre-arm was 'a confirmed lower high' and the letter is met (5.595 < 5.625) but the structure is not - an outside range that closed near its high, at the highest close of the sequence, on rising volume (160.8m vs 125.5m), with flat conviction, is accumulation not a failing high; (2) results land tonight with the reaction on 31 Aug, which is the date the 08-26 entry already designated for re-adjudication - trimming now IS pre-trimming into the catalyst; (3) optional_trim counterfactual n=58, median +1.72% saved at 1wk. G1 gave no cover (cv64), so the hold rests on the catalyst and the structure, not on G1. Trail yields nothing (chandelier 5.372 close-based / 5.387 high-based, both below the 5.44 cost-equal stop). RE-ARM SHARPENED - the bare 'lower high' hatch produced its first false positive today and is retired: 30% now fires at the 31 Aug EOD on ANY of a lower high WITH a lower close, a close below 5.44, or conviction below 55. · news |
| 3328 | Bank of Communications | 7.51 | NO ACTION | 7.51 | 7.42 | 7.48 (-0.4%) | 7.52 (+0.1%) | 7.68 (+2.3%) | Bank of Communications weakened - conviction fell 22 points, the close was down 0.79%, and the day printed both a lower high and a lower low ahead of interim results tonight. But the position is essentially at cost (+0.13%, or HKon), so the trim the model wants would bank roughly HKand is a stop-loss wearing a profit-take label. With the shares still above their stop, still trending, and a 12-week view 2.3% above the price, the position is held into the print and the stop is left to define the risk. Operational nuance: ADR-0012 tick (b) fires cleanly (-22) but HF-6 is decisive - a profit-take is unavailable absent a profit (Gain_ATR 0.07). No other origination class exists: Action HOLD, Consolidation_Bullish, close above stop, Target_12Week 7.68 non-inverted, so L2 fires on neither leg. Tick (c) does NOT fire (7.68 > 1.02 x 7.51 = 7.660). RISK STATED: the stop cushion has halved to HK = 0.63 ATR, the thinnest of the four mainland banks going into Monday's results reaction. Trail yields nothing (chandelier 7.224 close-based / 7.304 high-based, both below 7.42). RE-ARM: 30% fires on a decisive close below 7.42, or at the 31 Aug EOD if the results reaction leaves conviction below 55 with a lower close. · news |
| 2388 | BOC Hong Kong Holdings | 50.95 | NO ACTION | 50.95 | 49.58 | 51.07 (+0.2%) | 51.49 (+1.1%) | 52.20 (+2.5%) | BOC Hong Kong was the highest-conviction buy candidate in today's scan and it is not being bought, because the fundamental quality screen fails - the balance-sheet and cash-flow checks score 4 out of 9, and this book blocks new positions in names that fail that screen even when the chart is good. Everything else about the setup was acceptable: the entry sits at the market rather than above it, and the 1-week and 12-week views both point higher. Operational nuance: adjusted conviction 71 = raw 72 + val_adj -1 (OVERVALUED_VS_RI, MoS -0.271) - six points clear of the 65 floor. Blocked solely by the hard Quality_Pass=False gate, which applies to new entries only (2388 is unheld). HF-3 clears (T12 52.20 = +2.45%, non-inverted); HF-4 clears (Est_1W 51.07 > Entry 50.95); no-chase clears (Entry_Distance_ATR -0.0). Fourth consecutive session blocked, and the tally has turned: 51.00 -> 50.40 -> 50.00 -> 50.95 = -0.10% over the window, so the gate is no longer saving money while it has now prevented the book's single highest-conviction buy. See Systemic #5 - proposal is to make the gate advisory (a conviction haircut) rather than hard on books where >50% of rows are NA_*. Not news-assessed this run (blocked before the shortlist). · news |
| 0388 | HKEX | 424.00 | NO ACTION | 424.32 | 409.70 | 423.57 (-0.2%) | 423.82 (-0.1%) | 428.89 (+1.1%) | HKEX recovered strongly, closing up 0.76% and reclaiming a breakout label after two weak sessions, but it is priced well above what the earnings stream supports and the conviction score does not clear the bar once that valuation discount is applied. It also fails the fundamental quality screen, which independently blocks a new position. A better-priced entry or a stronger signal would change the answer. Operational nuance: adjusted conviction 58 = raw 64 + val_adj -6 (clamp(round(-1.186 x 5))) - seven points under the 65 floor - and Quality_Pass=False blocks the new entry independently. Entry_Distance_ATR 0.06 makes 424.32 a buy-stop just above the close. WATCH TRIGGER: raw conviction >=71 makes the valuation haircut alone non-binding; the quality gate would still need to clear or be downgraded to advisory per Systemic #5. · news |
Outcome & track record
Accuracy at the time of this record.
16 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 →
1.330 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.