Recommendations
1 to acquire · 2 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 |
|---|---|---|---|---|---|---|---|---|---|
| 1810 | XIAOMI CORP-W | 28.46 | BUY | 28.46 | 27.80 | 28.10 (-1.3%) | 26.68 (-6.3%) | 23.89 (-16.1%) | Xiaomi is the book's highest-ranked buy and one of only seven names in the universe that passes our fundamental quality screen. The buy signal has strengthened for five straight readings, the shares sit right at the model's entry price rather than extended above it, and momentum is moderate rather than overheated - so we are not chasing. We are taking a deliberately half-size, tightly stopped position: the risk is capped at about 2.3% of the position, because the model's own four- and twelve-week views are negative and our cash-flow valuation sits well below today's price. This is a tactical swing to the 30.90 objective, not a long-term holding. OPERATIONAL: Quality_Pass True (F_Score 6) - gate passes. Raw cv76 (Buy_Rank 1), val_adj -1 (MoS_FV -0.191), scorecard -2 (FailedBreakdown_Up_20D|buy n=12 hit_1w.417) -> adj 73 >= 65 floor. Entry_Distance_ATR 0.00, RSI 56.5, VWAP +2.40%, above EMA20 27.41 / EMA50 27.33 - all three HK-1 confirmation conditions met (RSI<70, multi-print, VWAP reclaim). SIZE DEVIATION STATED: model-faithful = 70.59% x HKx 20% = HK (97% of); capped at HK -> -> 14,000 on Xiaomi's 200-sh board lot = HK (7.3% of book, ~50% of). Grounded in HK-1 (never up-size a momentum-chase family) and HK-9's buy-side mirror. HARD STOP 27.80 = max loss HK = 0.17% of book; target 30.90 = HK; R/R 3.70:1. COUNTER-CASE CARRIED: all three Est_* horizons negative (-1.26% / -6.26% / -16.06%) and DCF_per_share 17.52 is 38% BELOW the price with Rev_Growth 25% - reason for half size and a hard stop, and the reason this must not be carried as a thesis position. Est_Model_Ver hk/20260703T082707Z(stale), Est_*_Conf 80* nominal - directional only. WARNING: Target_Price 30.90 (+8.57%) contradicts Est_4W_Point 26.68 (-6.26%) on the same row - see systemic finding 3. EXIT DISCIPLINE: close on 30.90 or 27.80; do not widen the stop. · news |
| 2513 | Z.AI (KNOWLEDGE ATLAS) | 1030.00 | SELL 30% | 1030.00 | 975.00 | 1056.17 (+2.5%) | 1080.18 (+4.9%) | 1211.12 (+17.6%) | This is the book's second-largest holding at roughly 19% of assets and it is down 37% from our cost. The shares have made a clear sequence of lower highs since July and now trade below both their one-month and two-month trend lines. We are taking only the minimum 30% off - keeping - because the stock joins a major global index at the end of this month, which usually brings buying. This trims a concentrated losing position without abandoning the index event. OPERATIONAL: THIS IS THE SINGLE PERMITTED HK-11 ESCALATION AND IT TERMINATES THE THREAD. Chain originated 08-21 on a fresh STRONG SELL - REDUCE cv73, Sell_Rank 1, FailedBreakout_Down_20D (best-sampled cell n=64, hit_1w.532, bias +1); re-issued 08-25; unexecuted both times (08-25 non-execution cost HK - the 08-26 open printed 1,072.00). The 08-25 card pre-committed in writing: 'the next EOD that still finds this unexecuted must escalate it once as market-at-open and then retire it to a numeric WATCH line.' Honoured. TENSION STATED: today's label has degraded to AVOID cv51, Sell_Rank blank, 0.0 - this card EXECUTES an origination already made, it does not create a new one. Structure still bearish independent of the label: confirmed lower-high -> -> 1030, today's high 1,072 < yesterday's, below EMA20 1,131.76 / EMA50 1,228.80, VWAP -6.75%, Pillar_Consensus Consolidation_Bearish. REPRICED from 1,059.00 per the provenance rule. Tier held at the 30% floor under the catalyst-aware-hold clause (MSCI ACWI effective at the 31-Aug close, direction up, stance hold-through - a trim is permitted on a confirmed lower-high, which is confirmed, but not a dump into a confirmed index add) plus HK-9 (hk sell signals are a one-week instrument). Cuts the line from 18.8% to ~13.2% of assets. NOT a profit-take (Gain_ATR -5.27, no PROFIT_ARMED). Model Stop_Price 1,030.00 prints AT the close - zero-width, rejected (HK-12), 2nd consecutive session; residual 700 trails 975.00 (-0.48xATR). All three Est_* horizons positive (+2.54% / +4.87% / +17.58%) - the honest case against acting, discounted because Est_Model_Ver prints (stale) and calib_1w is -0.261 (L5/HK-6). AFTER THIS CARD THE THREAD IS RETIRED AND MUST NOT BE RE-ISSUED, FILL OR NO FILL. RE-ARM: decisive settled close < HK. STAND DOWN: reclaim > EMA20 HK. · news |
| 3750 | CATL | 618.00 | SELL 33.3% | 618.00 | 600.00 | 615.88 (-0.3%) | 616.64 (-0.2%) | 634.36 (+2.6%) | CATL has now closed below the stop level we committed to for a second straight session, and the signal engine has been unable to form a view on it for twelve readings running - the trend indicators are in outright conflict. The shares sit below both their one-month and two-month averages. We are taking the minimum third off to reduce the bleed while keeping two-thirds of the position, because the twelve-week outlook is still mildly positive and the break is shallow rather than decisive. OPERATIONAL: exit basis is (i) the pre-committed 08-21 trail 'close < 625.50' broke on 08-25 and has NOT been reclaimed - close 618.00 is -1.20% / -0.37xATR below it, a 2nd consecutive settled close beneath the line; and (ii) CLAUDE.md SS9 'sustained NoTrade/Conflicted on a held position' - twelve consecutive NO TRADE prints across 08-25/08-26 (six today) at cv0, Pillar_Consensus Indeterminate, NoTrade_Reason verbatim 'Consensus conflict + low conviction'. Below EMA20 636.59 / EMA50 643.93, VWAP -3.53%, above EMA200 594.45. L2 does NOT fire (Est_12W 634.36 = +2.65%, flat-positive) so only the minimum tier is defensible; 1wk/4wk are flat-negative (-0.34% / -0.22%), which is what makes the trim worth taking rather than nothing. Valuation FAIR_VALUE MoS -1.045 sits in the neutral band -> NO tier bump (needs <= -3), no down-tier; it correctly originated nothing. DCF 723.08 > Target_12Week 634.36 -> no DCF-exceedance annotation. BOARD-LOT CAVEAT: 30% is but no HK board-lot table exists in the codebase (swingtrade/trading/engine.py:36 has only us/sg), so the card takes = 33.3%, stated rather than silently rounded. NOT a profit-take - PROFIT_ARMED is SPURIOUS (Gain_ATR -6.69 on a -17.8% position; arms off T12_Progress 97.42 alone). Position x 618 = HK3.4% of book. Residual 200 trails 600.00 (-0.90xATR, below today's 616.50 low), carried unchanged from 08-25. HK-11: RE-ISSUE # - the next EOD that still finds this unexecuted must escalate once as market-at-open and then retire it to a numeric WATCH line. 08-25 non-execution cost HK (the 08-26 open printed 621.00 exactly). · news |
| 0005 | HSBC HDGS | 161.70 | HOLD | 161.70 | 159.50 | 161.67 (-0.0%) | 161.35 (-0.2%) | 160.40 (-0.8%) | HSBC is the book's anchor at 35.5% of assets and continues to hold above our stop with a healthy long-term uptrend - it trades 17% above its 200-day average. The signal is a mild pullback-buy, our historically most reliable buy pattern, but the engine sizes no add, so we simply hold and keep the stop where it is. OPERATIONAL: HOLD cv63 (eased through the session) on PullbackBuy_VWAP - the book's BEST buy cell (bias +2, n=13, hit_1w.615, HK-2's positive anchor) - but 0.0 and no Buy_Rank, so there is NO engine-sized add and none is invented. Close 161.70 at EMA20 161.57, above EMA50 157.24, +17.0% above EMA200 138.17, VWAP -0.96%. PROFIT_ARMED (T12Prog 100.81) with tick (c) firing (Est_12W 160.40 < 1.02 x 161.70 = 164.93) but Avg_Cost is BLANK -> ADR-0012's no-guess rule blocks the profit-take; resolves to a trail action. Model Stop_Price 158.61 is BELOW the standing 159.50 line -> NOT lowered. Position = HKLargest rate-event exposure in the book. |
| 0522 | ASMPT | 162.90 | HOLD | 162.90 | 155.02 | 164.58 (+1.0%) | 172.07 (+5.6%) | 167.18 (+2.6%) | ASMPT flashed a sell signal in the middle of the session but it faded to no-view by the close, and our rule is that the settled close is what counts. The shares remain 5% above our stop. We hold and reassess after tonight's NVIDIA results, which are the single biggest swing factor for this name. OPERATIONAL: printed STRONG SELL - REDUCE cv70 TWICE (11:30 and 14:00) then faded to NO TRADE cv0 at both 15:30 and the settled close - NoTrade_Reason verbatim 'Consensus conflict + low conviction', Pillar_Consensus Indeterminate. L4 is explicit that the settled close is authoritative, so NO card is written. Close 162.90 is 5.1% above the trail and +1.41% vs VWAP, but below EMA20 165.25 / EMA50 169.01 after a wide 159.00-169.90 range. PROFIT_ARMED SPURIOUS (-17.04%, Gain_ATR -3.45). Position = HK3.3% of book. Trail held 155.02. |
| 0700 | TENCENT | 445.40 | HOLD | 445.40 | 430.00 | 441.19 (-0.9%) | 436.40 (-2.0%) | 446.03 (+0.1%) | Tencent prints the book's second-strongest sell signal for a sixth straight session, but we retired this exit on 17 August to a specific price line - HK - and the shares are still 3.6% above it. The position is only 1.6% of the book, so overriding that discipline for a HKtrade is not worth it. We hold and let the line do the work. OPERATIONAL: STRONG SELL - REDUCE cv73, Sell_Rank 2, 85.46, adj 74 (bias +1) - deliberately NOT acted on; HK-11 binds. Exit retired 08-17 to the 430.00 line, which is ALSO the news overlay's standing per-ticker override ('close below 430 -> exit'). Model Stop_Price 423.28 is BELOW 430.00 -> not lowered. Below EMA200 498.56, VWAP -3.52%. Horizons -0.94% / -2.02% / +0.14%. Avg_Cost = 1.00 is a DATA ERROR (Unreal_PnL_Pct +44,440%, Gain_ATR 40.18), unchanged since 08-18 and now 7 sessions old - PROFIT_ARMED is meaningless on this row. Position = HK1.6% of book. |
| 0823 | LINK REIT | 39.34 | HOLD | 39.34 | 38.70 | 39.42 (+0.2%) | 39.38 (+0.1%) | 38.95 (-1.0%) | Link REIT keeps strengthening - conviction climbed through the session to 75 and the shares hold above all three of their trend averages. We withdrew a buy order on this name yesterday when conviction collapsed, and we are not re-queuing it: the stock is now at the top of its range and chasing it would be a worse entry. We hold what we have. OPERATIONAL: HOLD cv75 -> ->. Close 39.34 above EMA20 39.00 / EMA50 38.71 / EMA200 38.32, VWAP +0.91%. 0.0 and no Buy_Rank -> no add proposed. The withdrawn 08-21 GTC buy-stop @39.81 is NOT re-queued - the provenance rule forbids republishing a stale limit, and the live row prints a Market entry at 39.34. PROFIT_ARMED (T12Prog 101.00) with tick (c) firing (Est_12W 38.95 < 1.02 x 39.34 = 40.13) but Avg_Cost BLANK -> no-guess blocks it. Model Stop_Price 38.32 is below the standing line -> trail held 38.70. Position = HK2.0% of book. |
| 0939 | CCB - H SHS | 9.22 | HOLD | 9.21 | 8.88 | 9.16 (-0.6%) | 9.31 (+1.0%) | 9.35 (+1.5%) | CCB reports half-year results with an interim dividend on 28 August - the only scheduled earnings event among our holdings in the next two weeks. The signal faded to no-view into the close, but the shares are 10% above their long-term average and we do not trim into a dividend print. Hold through the results. OPERATIONAL: faded HOLD 58/58/58/61 -> NO TRADE cv0 at the settled close, Pillar_Consensus CONFLICTED, NoTrade_Reason verbatim 'Consensus conflict + low conviction'. Close 9.215, +10.5% above EMA200 8.34, +1.98% vs VWAP, RSI 63.0. CATALYST-AWARE HOLD applies: interim results + interim dividend 28-Aug after the HK close (horizon stance tighten-trail). PROFIT_ARMED (T12Prog 98.56) with tick (c) firing (Est_12W 9.35 < 1.02 x 9.215 = 9.40) but Avg_Cost BLANK -> no-guess blocks it. Position = HK8.5% of book. Trail held 8.88. |
| 0981 | SMIC | 69.70 | HOLD | 69.70 | 69.63 | 70.08 (+0.5%) | 72.85 (+4.5%) | 76.22 (+9.4%) | SMIC's stop breach has healed - the shares closed back above our HKline and reclaimed their 200-day average. But NVIDIA reports tonight and Hong Kong trades that news tomorrow, which is exactly the event we deferred this decision for. The line stays live: any close back below HKtriggers an automatic 30% trim at the following open, with no further debate. OPERATIONAL: BREACH HEALED - close 69.70 vs the 69.63 trail (+0.10%) after a 68.15-71.50 range; back above EMA200 68.90, still below EMA20 69.96 / EMA50 71.27. The 08-25 stand-down condition ('stand down only on a settled close back at or above 69.63') is satisfied VERBATIM. BINDING AND CARRIED FORWARD UNCHANGED: a settled close below HKon 27-Aug or any session after -> MANDATORY 30% trim at the next open, no re-evaluation, no lowering of the line. Signal AVOID cv63 with 0.0 and no Sell_Rank -> no engine-sized sell today. Model Stop_Price 77.41 prints 11.1% ABOVE the close - INVERTED, rejected (HK-12), 2nd consecutive session. PROFIT_ARMED but Avg_Cost BLANK -> no-guess blocks ADR-0012. Position = HK3.2% of book. Trail held 69.63 - NOT re-based in either direction. |
| 1347 | Hua Hong Semiconductor | 117.50 | HOLD | 117.50 | 105.00 | 121.94 (+3.8%) | 132.00 (+12.3%) | 137.22 (+16.8%) | Hua Hong prints the strongest sell signal in the book today - but we retired this exit on 19 August after three unexecuted orders, and set two explicit price triggers: sell below HKstand down above HKThe shares closed at HK, between the two, after a 6.4% up day. Selling into that rally would repeat exactly the mistake that retirement was designed to stop. The stock also joins the Hang Seng Index in early September, which typically brings index buying. We hold. OPERATIONAL: STRONG SELL - REDUCE cv76, Sell_Rank 1, 85.46, strengthening all session (AVOID53 -> -> -> 76), adj 77 (bias +1), and MoS_FV -19.77 would bump a trim one tier IF one were live. IT IS NOT. The 08-19 HK-11 retirement set RE-ARM: decisive settled close < HK; STAND DOWN: reclaim > HK - close 117.50 sits BETWEEN them and NEITHER has fired. Today was +6.43% (110.40 open, 110.00 low, 119.80 high). All three horizons strongly positive (+3.78% / +12.34% / +16.78%). Below EMA20 128.60 / EMA50 139.05, above EMA200 115.80, VWAP -9.38% (widest dislocation in the book). Model Stop_Price 117.49 against a 117.50 close - ZERO-WIDTH, rejected (HK-12). Position = HK6.4% of book, -26.6% vs Avg_Cost 160.00. Trail held 105.00. |
| 2888 | STANCHART | 229.60 | HOLD | 229.60 | 229.36 | 229.92 (+0.1%) | 230.92 (+0.6%) | 220.22 (-4.1%) | Standard Chartered's conviction collapsed from through the session and the close sits just 0.1% above our stop - the tightest line in the book. On a bigger position we would already be trimming, but this is 1.5% of assets and we do not have a recorded, which blocks our profit-taking rule. We are pre-committing instead: any close below HKtriggers a 30% trim at the next open. OPERATIONAL: conviction -> ->, a -32 pt drop that IS ADR-0012 fire tick (b) on an armed row - and the row IS PROFIT_ARMED (T12Prog 104.26). BLOCKED because Avg_Cost is BLANK (5th consecutive session the no-guess rule has had to fire; and the FIRST time it has blocked a conviction-collapse fire rather than a 12wk-view fire - systemic finding 4). Close 229.60 below EMA20 232.31 and VWAP (-1.57%), still 17.3% above EMA200 195.77. Model Stop_Price 218.54 is below the standing line -> NOT lowered; trail held 229.36. PRE-COMMITTED: a settled close below 229.36 arms a 30% trim at the next open. Position = HK1.5% of book. |
| 7709 | XL2CSOPHYNIX | 37.50 | HOLD | 37.50 | 28.00 | 38.70 (+3.2%) | 37.26 (-0.6%) | 38.60 (+2.9%) | This leveraged ETF stays retired for a ninth session. The sell label actually weakened today, the shares are 34% above our re-entry-to-sell line, and our standing rule for 2x leveraged products requires far stronger evidence before a full exit. No action. OPERATIONAL: AVOID cv50 (the sell label WEAKENED from SSR39), 0.0, no Sell_Rank. Re-arm is 'decisive close < 28.00 OR a breakdown-family signal' - close 37.50 is 34% above the line and the family is failed_breakout, not breakdown. The leveraged-2x rule (memory/feedback_leveraged_etf_exits.md) independently blocks a full exit: it requires conv >= 70 PLUS 3 prints PLUS price < stop; cv50 is nowhere near. HK-12 REGRESSION TO RECORD: Stop_Price prints 46.90, i.e. 25% ABOVE the 37.50 close - INVERTED AGAIN after correcting to a valid 26.04 on 08-25. PROFIT_ARMED SPURIOUS (-67.3%, Gain_ATR -16.39). Position = HK1.2% of book. Trail held 28.00. |
Outcome & track record
Accuracy at the time of this record.
181 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 →
-0.261 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.
49 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.