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 | 229.60 | HOLD | 229.60 | 228.47 | 229.92 (+0.1%) | 230.92 (+0.6%) | 220.22 (-4.1%) | Standard Chartered is held rather than trimmed even though it just posted the book's sharpest one-day setback: the shares fell 2.88% to HKand the model's conviction in the setup collapsed from, wiping out nearly all of an unrealised gain that stood at +3.40% a day earlier and is now +0.42%. There is simply no profit left to bank - a 'profit-take' here would realise about HKand is really a stop-loss under another name - and the bank's own trading signal has not turned bearish: the engine still reads the setup as a bullish consolidation and the price is still above our protective sell level. Rather than sell into a single bad session with no news behind it, we hold and let a pre-set exit make the decision. OPERATIONAL: ADR-0012 fires on paper with ticks (b) conviction -31pts and (c) Target_12Week 220.22 < 1.02x229.60; both suppressed under HF-2 - the PROFIT_ARMED flag is an inverted-12wk artifact (T12_Progress 104.26 is produced by T12 < Close) and tick (c) is that same inverted fact read twice. HF-2's independent bearish basis (decisive close < stop, or fresh high-conviction formal SELL) is absent: Action HOLD, Pillar_Consensus Consolidation_Bullish, close 229.60 > stop 228.47. Gain_ATR 0.18 vs the 3.0 winner bar; optional_trim counterfactual n=56 median +1.72%/1wk saved. Stop unchanged at 228.47 (peak stop; cushion HK = 0.204 ATR, thinnest in the book, and HKbelow cost so a break is a break-even exit). Trail-raise yields nothing and is NOT adopted: 2.0xATR chandelier off close = 218.54 (the engine's own printed Stop_Price, down from 226.48 as ATR widened) and off the 234.60 high = 223.54 - both below the recorded stop; stops never move down. PRE-ARMED: a decisive close below 228.47 is a FULL EXIT, not a trim - L2 (stop break + negative 12wk, Est_12W -4.11%) and HF-2's independent basis fire together. News none found (Yahoo 503; a Seeking Alpha Goldman upgrade item returned 403, date unverifiable, NOT applied). UK-domiciled, outside the 28 Aug mainland cluster, exposed to Jackson Hole 28 Aug and US PCE 26 Aug. · news |
| 3988 | Bank of China | 5.55 | HOLD | 5.55 | 5.44 | 5.53 (-0.5%) | 5.63 (+1.4%) | 5.67 (+2.1%) | Bank of China is the book's strongest holding, up 2.11% on the position and grinding to successively higher highs and higher closes, and it publishes first-half 2026 results in two trading days. A momentum indicator has flagged a possible loss of upward thrust, which would ordinarily trigger a partial sale to lock in gains - but selling immediately before a scheduled earnings release means guessing at the result rather than reading it, and every observable price fact still points up. We hold the full position through the announcement and keep a protective sell level 2.1% below the market. OPERATIONAL: ADR-0012 arm is LEGITIMATE, not an artifact - Target_12Week 5.67 > Close 5.555 (non-inverted), T12_Progress 97.97, Gain_ATR 1.32 (highest in the book). Tick (a) RSI_DIV fires -> mandatory 30% trim = (~HK). G1 offers no cover (cv64, not a fresh cv>=80 breakout), so the suppression rests ENTIRELY on the catalyst-aware hold: board meets 28 Aug 2026 to approve 1H results and consider an interim dividend (confirmed). Neither escape hatch met - higher high 5.625 > 5.595, higher close, stop intact at 1.32 ATR. Trail-raise yields nothing (2.0xATR chandelier 5.380 < 5.44 = cost). Stop unchanged 5.44 (= peak stop = cost). RE-ARM: 30% fires immediately on a confirmed lower high or a close below 5.44; re-adjudicate at the first EOD after the 28 Aug print. DATA FLAG (3rd consecutive session): RSI_DIV is unobservable in the 8-session window - RSI has risen 63.7 -> 65.0 -> 65.5 into higher highs and higher closes. This single unverifiable print is the sole trigger for a 6,000-share trim; see systemic finding #3. · news |
| 3328 | Bank of Communications | 7.57 | HOLD | 7.57 | 7.42 | 7.53 (-0.5%) | 7.57 (+0.0%) | 7.68 (+1.5%) | Bank of Communications is strengthening, not weakening - the model's conviction in the setup rebounded from, the shares touched their highest level since we bought them, and the position is up 0.93%. A mechanical rule nonetheless flagged a partial sale, purely because the twelve-week price projection sits a fraction below a 2% hurdle. That is noise, not deterioration, and the bank reports first-half results in two trading days alongside a decision on an interim dividend. We hold the full position into the announcement. OPERATIONAL: arm is legitimate (Target_12Week 7.68 > Close 7.57, T12_Progress 98.57, Gain_ATR 0.56). Tick (c) fires by 0.55% - 7.68 < 1.02 x 7.57 = 7.7214 -> 30% trim = -> at board lot 1,000. Tick (b) does NOT fire; conviction ROSE +15, Momentum, and today's 7.665 high is the highest print since the position opened. SUPPRESSED on the catalyst-aware hold (1H results + interim-dividend board meeting confirmed 28 Aug, two sessions out) with both escape hatches unmet - higher high 7.665 > 7.615, higher close, stop intact at 1.20 ATR. Trail-raise yields nothing (chandelier 7.320 < 7.42). Stop unchanged 7.42 (= peak stop). The 08-21 GTC limit @7.53 stays WITHDRAWN and is NOT re-posted: the row prints HOLD with 0.0, so there is no live entry trigger to anchor a limit to. RE-ARM: 30% on a confirmed lower high or a close below 7.42; re-adjudicate after 28 Aug. This card is the cleanest example of systemic finding #1 - a trim triggered on a strengthening name by a near-flat 12-week projection. · news |
| 3968 | China Merchants Bank | 49.42 | HOLD | 49.42 | 48.10 | 49.54 (+0.2%) | 49.75 (+0.7%) | 50.04 (+1.3%) | China Merchants Bank is held in full. The model's conviction in the setup improved from and the position is up 1.58%, yet a mechanical rule again suggested trimming - for the third session running - because the twelve-week projection sits marginally below a 2% hurdle. The projection and the trigger are the same number counted twice, so acting on it would be selling a strengthening position on a technicality. Unlike the book's three mainland state banks, no results date could be confirmed for this name, so it carries an unscheduled-announcement risk we are watching rather than pre-empting. OPERATIONAL: arm non-inverted (Target_12Week 50.04 > Close 49.42, T12_Progress 98.76, Gain_ATR 0.88). Tick (c) fires (50.04 < 1.02 x 49.42 = 50.4084) -> 30% =. Tick (b) does NOT fire; conviction rose +11. SUPPRESSED on the HF-2 mechanism, consistent with 08-21 and 08-25: arm and tick (c) are one T12 relation read twice and Gain_ATR 0.88 << 3.0 gives no independent winner basis. Flipping to a trim on identical-but-improved facts after declining it twice would be arbitrary. Stop unchanged 48.10 (= peak stop, cushion 1.51 ATR). Trail-raise yields nothing on the book's established CLOSE-based method: chandelier 47.67 (= the engine's own printed stop) < 48.10. NOTED, NOT APPLIED (3rd session): the HIGH-based chandelier 49.90 - 1.754 = 48.15 would permit a HKraise; switching basis mid-position is an undisclosed methodology change - see systemic finding #4. CALENDAR GAP: 1H results date not confirmable on the permitted domains and the name has no Event Horizon coverage (parent hk_horizon.json covers only 0005/0939/2888) - the one holding with neither catalyst protection nor catalyst warning. · news |
| 0939 | CCB - H SHS | 9.22 | HOLD | 9.21 | 8.89 | 9.16 (-0.6%) | 9.31 (+1.0%) | 9.35 (+1.5%) | China Construction Bank is held in full into its first-half results, published after the Hong Kong close on 28 August with the market reaction landing 31 August. The trading model has stopped producing a reading on this name for two sessions running - its internal indicators disagree with each other - and a mechanical rule reads that silence as a reason to sell. Our own track record says otherwise: every previous time this book sold into that particular condition it was wrong, and the position has just crossed back into profit while making a higher high and a higher close. We hold, with a protective sell level 3.5% below the market, and accept openly that a sharp adverse reaction to the results could gap straight through it. OPERATIONAL: 2nd consecutive NO TRADE cv0 (Pillar_Consensus Conflicted, Momentum 24, NoTrade_Reason 'Consensus conflict + low conviction') means tick (d) GENUINELY fires on its own merits for the first time, joined by tick (c) (Target_12Week 9.35 < 1.02 x 9.215 = 9.3993) - two ticks = profit_take_2 = 50% = on paper. SUPPRESSED on three grounds: (1) no profit to take - Gain_ATR 0.09, +HK (+0.16%), just across breakeven from -0.22%, so HF-6's principle applies in substance; (2) DECISIVE - scorecard cell NoTrade|sell is ACTIONABLE at bias -10 with hit_1w 0.0 on n=5, and tick (d) IS selling on NoTrade, so firing it would knowingly repeat this book's worst-graded behaviour; (3) catalyst-aware hold, with both escape hatches unmet - higher high 9.345 > 9.325 and higher close. Stop unchanged 8.89 (= peak stop, 1.95 ATR, the book's widest). Trail yields nothing (chandelier 8.882 < 8.89; the NoTrade row prints no stop at all). RISK ACCEPTED AND SIZED: stop sits 3.53% below the close against a +/-4% Event Horizon results band - HKto stop on a HKposition, and the stop will not save this line on 31 Aug. Re-adjudicate at the first EOD after the print. · news |
| 0005 | HSBC HDGS(USD0.50) HK O'SEAS | 161.70 | NO ACTION | 163.33 | 160.77 | 161.67 (-1.0%) | 161.35 (-1.2%) | 160.40 (-1.8%) | HSBC is not bought. It ranks first in the book on raw screening score, but the model's conviction in the setup has fallen from - below the threshold at which this book commits capital - after the shares dropped 0.92%. The model's own twelve-week and four-week price projections both sit BELOW today's close, so on its own numbers the trade has no forward return, and the suggested entry price is HKABOVE where the stock just closed: that is paying up into a falling name, the precise pattern this book has learned to avoid. This is the fourth consecutive session we have looked at HSBC and declined; over that span it has drifted from HKto HK, so the patience has cost nothing. OPERATIONAL: adj_conviction 56 vs the 65 floor - rejected on the floor alone, no further test needed. Signal flipped FailedBreakdown_Up_20D -> PullbackBuy_VWAP (n=1, report-only, bias +0); val_adj 0 (NA_CCY, gate off). Three independent legs would also reject: HF-3 (Est_12W 160.40 < Close 161.70, inverted 2nd session), HF-4 (Est_1W 161.67 < Entry 163.33 - the estimation layer contradicts the conviction layer), and Entry_Distance_ATR 0.65 making it a buy-STOP into a down day (L1 chase). Momentum. History: queued 08-17 @163.50, withdrawn 08-18, rejected 08-21 and 08-25. Ledgered optional:true for counterfactual grading. RE-ARM: a BUY at raw conviction >=75 (the -10 FailedBreakdown_Up_20D|buy haircut applies if it reverts to that family) with Est_12W above the close. · 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 →
1.047 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.