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 |
|---|---|---|---|---|---|---|---|---|---|
| 2888 | STANCHART | 241.80 | SELL 46.7% | 241.80 | 233.48 | 239.33 (-1.0%) | 237.73 (-1.7%) | 234.89 (-2.9%) | Standard Chartered has run 5.8% above our cost and the trade is now going the other way: conviction collapsed from in a single session as the 52-week breakout signal degraded, and the model's own one-week, four-week and twelve-week price paths all now sit BELOW today's close (-1.0%, -1.7%, -2.9%) - the only holding in the book where every horizon points down. We bank a little under half the position at 241.80 and trail the rest, with the Fed and the HKMA both moving on 17 September in a direction that squeezes Hong Kong bank margins. Profit-take: armed (T12_Progress 102.94) + 2 ticks - (b) conviction -35pts, (c) Est_12W -2.86%. HF-2 suppression (Gain_ATR 2.96 < 3.0 with an inverted 12wk) OVERRIDDEN by tick (b) as the required independent bearish basis. 750 x 50% = 375 -> lot 50 -> 7.5-lot tie -> rounded DOWN to 350 per the n=89 optional_trim counterfactual. Supersedes the UNFILLED 09-09 card (SELL 200 @242.60); destination is on either execution path - if the executor books that 200 late for execute_on 2026-09-10, sell 150 here instead of 350. Stop HELD 233.48 by ratchet (model 232.90 / chandelier 232.91 both lower). Valuation NA_CCY -> val_adj 0, no tier bump, quality gate off. · news |
| 3968 | China Merchants Bank | 52.15 | SELL 33.3% | 52.15 | 50.88 | 51.92 (-0.4%) | 52.38 (+0.4%) | 51.93 (-0.4%) | China Merchants Bank is up 7.2% on our cost and has now travelled more than three times its average daily range above where we bought it - far enough that the position qualifies as a winner worth protecting rather than pressing. The twelve-week view has gone slightly negative while the four-week is still marginally positive, so we take a third off at 52.15 into strength and let the remaining run behind a 50.88 stop, ahead of the 17 September Fed and HKMA decisions that compress bank margins. Profit-take: armed - Gain_ATR CROSSED 3.0 today (2.83 -> 3.22), so the arm is genuine on its own and HF-2 suppression LIFTS (same crossing that lifted 2888 yesterday). 1 tick: (c) Est_12W -0.42%. Disclosed weakness - Est_4W is +0.44%, so only the 12wk is inverted, by four tenths of a percent; the card ships on the arm, not the tick. Volume: 1,500 x 30% = 450 against a 500-share board lot = 0.; rounding DOWN would delete a fired take entirely, so this rounds UP to 500 (33.3%), a 3.3pp deviation, disclosed rather than silently re-tiered. Stop HELD 50.88 by ratchet (model 49.98 / chandelier 49.98 lower). Valuation NA_CCY -> val_adj 0, quality gate off. Note: a search snippet quoting 45.92 for 3968 was DISCARDED - L4, the anchor CSV close of 52.15 is authoritative. · news |
| 0939 | CCB - H SHS | 9.77 | SELL 50% | 9.77 | 9.37 | 9.74 (-0.3%) | 9.90 (+1.3%) | 9.82 (+0.5%) | CCB has climbed to a 6.1% gain and RSI has pushed to 67 while the price momentum underneath it has started to lag - a fresh bearish divergence appeared today for the first time in this run, and the twelve-week projection has flattened out to roughly where the stock already trades. We take half off at 9.77 and keep running behind a 9.37 stop rather than give back a good move into next week's Fed and HKMA decisions. HONEST CAVEAT ON THIS CARD: conviction actually ROSE and both the four-week (+1.3%) and twelve-week (+0.5%) estimates are POSITIVE - on any read but the rulebook's this is a hold, and the n=89 optional_trim counterfactual (median +5.88% wh_4w saved) argues directly against it. Profit-take: armed genuinely (T12_Progress 99.44 on a NON-inverted Est_12W, so HF-2 does not apply) + 2 ticks - (a) fresh RSI_DIV, absent from all three prior snapshots, NOT shielded by G1 because the signal is FailedBreakdown_Up_20D rather than a fresh BreakoutUp_*/MomentumContinuation_*; (c) Est_12W below the 9.9603 threshold. 2 ticks -> profit_take_2 50% = = EXACTLY one board lot, so the tier and the lot agree for the first time and yesterday's purely-arithmetic deferral (30% = 600, un-expressible) no longer applies. Stop HELD 9.37 by ratchet (model 9.27 / chandelier 9.266 lower). Valuation NA_CCY -> val_adj 0. · news |
| 3328 | Bank of Communications | 8.06 | HOLD 1,500 | 8.11 | 7.72 | 8.02 (-1.1%) | 8.17 (+0.7%) | 8.17 (+0.7%) | Bank of Communications is the strongest chart in the book - a genuine momentum trend, top-ranked buy signal, 7.5% ahead of cost - and we are neither adding to it nor cutting it today. Not cutting, because trimming a strong uptrend on a stretched RSI alone is how you sell your best position too early; instead we raise the protective stop to 7.72. Not adding, because the model's own one-week price estimate sits below the price we would have to pay to get in, and next week's Fed and HKMA decisions argue for tightening stops rather than cash. Profit-take FIRED but held on BOTH axes: tick (a) RSI_DIV is G1-DOWNGRADED to a trail-raise (fresh MomentumContinuation_Up at cv 90, Tier-1 code - the NBIS protection); tick (c) clears its threshold by only 0.63% (8.17 vs 8.2212). 1 tick -> 30% = against a 1,000-share board lot -> minimum expressible cut 66.7%, two tiers worse, on a line already cut 40% effective 09-09. STOP RAISED 7.61 -> 7.72 (model stop; chandelier 7.673 and carried 7.61 both lower) - mandatory trail-raise carrying the downgraded RSI_DIV. · news |
| 3988 | Bank of China | 5.97 | HOLD 5,000 | 5.97 | 5.79 | 5.98 (+0.2%) | 5.98 (+0.2%) | 5.95 (-0.3%) | Bank of China is our best-performing holding at +9.7% and it does clear the bar for a partial profit-take, but the case for cutting it today is the thinnest of the four candidates - its twelve-week estimate is only a third of a percent below the price while its one-week and four-week views are both positive. With only three disposal slots and three stronger cases ahead of it, we hold the full behind a 5.79 stop. Profit-take armed most genuinely in the book (Gain_ATR 3.83, T12_Progress 100.34) but fires on 1 tick only - (c) Est_12W -0.34% while Est_1W and Est_4W are both +0.17%. Ranked 4th of 4 by dollar bleed (HKvs 2888's HK) against the 3-card cap. Its UNFILLED 09-09 card (SELL 1,000 @5.915) is withdrawn and reconciled separately, NOT re-issued; if the auto-executor books it late for execute_on 2026-09-10 that fill stands as a legitimate 09-09 decision and no further 3988 action is taken. Stop HELD 5.79 by ratchet (model 5.69 / chandelier 5.693 lower). · news |
| 3328 | Bank of Communications | 8.06 | NO ADD | 8.11 | 7.72 | 8.02 (-1.1%) | 8.17 (+0.7%) | 8.17 (+0.7%) | This was the strongest buy signal the book has produced - top-ranked, highest conviction on record - and we declined it, because the model's own one-week price forecast (8.02) sits below the 8.11 we would have to pay to get filled. When the forecast and the conviction score disagree that directly, the forecast has been right here before: the identical pattern on Standard Chartered in July cost 2.2% the very next session. RSI at 71 with a bearish divergence already printed on the same row, and a Fed/HKMA decision on 17 September, say wait for a better price rather than a better story. WITHHELD ACQUIRE - adj conviction 90 clears the >=65 floor by 25 points and PASSES HF-3's non-inversion gate (Est_12W +1.36%); rejected on HF-4 (LOW_QUALITY_ENTRY) alone, reinforced by HF-3's cell record in this book (MomentumContinuation_Up|buy 0-for-3, stop-first 1.0) and the horizon's tighten-trail stance into US CPI 09-14 / FOMC+HKMA 09-17. COUNTER-ARGUMENT DISCLOSED: no_chase_queue counterfactual n=9, median +4.97% wh_1w [cost] - declining these prints has cost this book nine times running, and the book is now 90% cash. RE-ARM TRIGGER: Est_1W_Point >= the printed Entry_Price on a subsequent print, AND after the 09-17 Fed/HKMA pair clears. · news |
Outcome & track record
Accuracy at the time of this record.
39 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.