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 |
|---|---|---|---|---|---|---|---|---|---|
| 0992 | Lenovo Group Limited | 29.04 | HOLD | 29.54 | 24.84 | 32.00 (+8.3%) | 38.92 (+31.8%) | 41.38 (+40.1%) | HOLD the book's only position through tomorrow's results. Lenovo is the HK-listed AI-server bellwether and the case strengthened on every layer today: the stock closed +1.75% at 29.04, printed 29.18 against a 52-week high of 29.38, conviction rose, and Morgan Stanley upgraded the name to Overweight while more than doubling its price target to HKfrom HK - explicitly arguing that the memory-cost inflation which was this position's stated bear case is being fully passed through to customers with margins preserved. The 12-week projection of 41.38 still sits inside the 43.49 DCF anchor and the stock trades far above every moving average. The engine is still not offering more shares - the label is HOLD with 0.0 and a re-arm entry of 29.54 sitting above the close - and we would decline an add regardless with a two-sided result landing tomorrow. OPERATIONAL: Q1 FY2026/27 results 2026-08-13, ONE session out, two-sided +/-6.0% band (close 27.30-30.78, position 273,000-307,800, a +/-0.86%-of-book swing). Catalyst-aware hold applies with maximum force - hold and trail THROUGH the print; a trim would require a confirmed lower-high or a decisive stop break and there is neither. L6 blocks any add before the print. Every sell-origination class is closed: no Exit_Warning code; NOT PROFIT_ARMED (Gain_ATR 0.99 needs 3.0, i.e. a close of 32.19; T12_Progress 70.18 needs 90, i.e. 37.24), so ADR-0012 cannot fire and its conviction-collapse / dead-12wk / repeated-NO-TRADE ticks are unreachable; no stop break (14.5% headroom); 12wk +42.5% so L2's second leg fails independently; label is not bearish. ADR-0020 trail erosion 0% - peak stop rose for a FIFTH straight session (21.31, 21.52, 22.73, 24.33, 24.84); the chandelier lifted the live stop 24.33 to 24.84, cutting risk-if-stopped to (27.498-24.84)x10,000 = HK = 1.32% of book. Position columns are CLEAN today (Avg_Cost 27.50, Gain_ATR 0.99) - the 08-11 staleness was a fill-day artifact of the scan-before-executor race, confirming Systemic #2. Valuation FAIR_VALUE, MoS -0.6056, val_adj -3, Quality_Pass True, F-Score 7, Quality_Rank; adjusted conviction 71 now clears the 65 floor where yesterday's 63 did not. FORWARD-NUMBER CAVEAT: target_4w 38.92 sits 18.2% above Est_High 32.92 (series 17.1, 17.1, 20.3, 18.2 - persistent, not monotonically widening; Systemic #3 revised); 32.92 is the honest upper end and MS's HKis 22.9% below the model objective. f=0.30 mandatory per L5 (calib_1w -0.346). HT-6 re-checked: no placement or registration - CLEAR. EXIT TRIGGERS: decisive close below 24.84; OR any Exit_Warning code; OR a SELL - REDUCE label; OR two consecutive NO TRADE/sub-40 prints with Target_12Week under the close (L2). RE-ARM FOR AN ACQUIRE: a BUY - ADD label with > 0, mechanically a close/trigger through 29.54, but not until the 08-13 result is out and digested. · news |
| 9618 | JD-SW | 123.00 | WATCH | 124.79 | 120.20 | 124.80 (+0.0%) | 128.99 (+3.4%) | 144.44 (+15.7%) | NO BUY on JD.com, and this is the closest this book has come to a genuine buy - worth being clear that it is a timing and plumbing refusal, not a rejection of the business. Everything a sound long needs is here: the e-commerce franchise is priced at fair value with a clean quality screen, the forward view is positive on both horizons (4-week 128.99 = +4.9%, 12-week 144.44 = +17.4% against a 123.00 close), the stock trades above its 50-, 100- and 200-day averages in a Trending regime with a Consolidation_Bullish pillar read, the stop at 120.20 is a real risk-definable level 2.3% below the market, and the entry at 124.79 sits ABOVE the close, so this would be an entry on strength rather than a chase. It is also PullbackBuy_VWAP, the family L1 favours. We are standing aside for three reasons, and the third is binding. (1) Conviction 62 is 3 points below the 65 floor, with val_adj 0 and scorecard_bias 0 there is nothing to lift it, and the floor is the engine's own BUY - ADD threshold - not to be re-invented downward for a name one likes. (2) R:R is 0.92 - risk 4.59 (124.79 to 120.20) against reward 4.20 (to 128.99) - which is HT-3 to the second decimal, the same config artifact measured on five names on 07-17 and on 9988 again today; the tuned HK pullback exits put the stop 1.25 ATR below entry against a 4wk target only ~1.145 ATR above, so this is a property of the config and not of JD, and it cannot be hand-fixed (the 1.25/2.5 re-pairing is gated behind the champion/challenger shadow test, >=20 sessions). (3) L6 - JD reports Q2 before the US open on 2026-08-13, i.e. after the HK close, so an order queued tonight fills at the 08-13 open and holds naked through the print with the reaction landing 08-14; the stock is already +22% over a month into an is-the-upside-already-priced-in setup. Consensus is +3.5% revenue and +24.6% EPS y/y. RECONCILIATION: the 08-11 watchlist trigger for this name (a BUY label with Target_Price ABOVE the close and R:R >= 1.5) HALF-fired today - the first clause fired for the first time, the second failed at 0.92 exactly as that trigger's own wording anticipated. RE-ARM: revisit AFTER the 08-13 print, queue-below only, and only with R:R >= 1.5 - which in practice means after HT-3's config artifact clears champion/challenger. Also logged: this row prints Order_Type 'Buy Limit @ HK' against a 123.00 close, a limit ABOVE the market that would fill instantly at the open when the intent is a buy-stop (Systemic #4). · news |
| 9988 | BABA-SW | 122.60 | WATCH | 119.76 | 113.90 | 119.36 (-0.3%) | 125.12 (+4.5%) | 137.26 (+14.6%) | NO BUY on Alibaba despite it printing the day's top-ranked buy signal. The setup is real - a pullback entry in the book's favoured family, the model's 12-week projection at 137.26 is 12% above the 122.60 close - but the fundamental screen fails and the price structure is a bear-market rally, so the book stays out. The blocking reason is absolute: Quality_Pass is FALSE on an F-Score of 3, and Alibaba is a zero-position new entry, which is exactly and only what the quality gate blocks (it never blocks adds to held names and never forces a sell). Three further independent rejections stack behind it, any one sufficient. Valuation is GROWTH_PRICED_IN at MoS -0.6873, so val_adj -3 takes conviction 64 to an adjusted 61, four points below the 65 floor. R:R is 0.91 - risk 5.86 (119.76 to 113.90) against reward 5.36 (to 125.12) - the HT-3 config artifact again, now measured on a fourth and fifth independent name. And the close of 122.60 sits BELOW the 200-day average at 127.24, which is precisely HT-4's bear-rally-in-disguise structure: a high-conviction pullback signal on a name under its long-term trend line. It is also extended 0.61 ATR PAST its own entry (close 122.60 vs entry 119.76), so the no-chase rule would demote it to a queue even had it cleared everything else. Trail context: WATCH 44, WATCH 53, WATCH 44, then BUY 64 - it entered the buy pool for the first time in this window. CALENDAR: Alibaba reports June-quarter results on 2026-08-20, six sessions out - outside the <=5 catalyst window but inside the holding period of any buy made this week, and NOT listed in the hk Event Horizon file (Systemic #8). RE-ARM: a Quality_Pass True print, or a reclaim of the 200-day average with conviction >= 68 and R:R >= 1.5 - and not before the 08-20 result. · news |
| — | AI - HK Tech | — | NO ACTION | — | — | n/a | n/a | n/a | NO ACTION both sides - an affirmative verdict. Unlike the last three sessions the buy pool was NOT empty: two names cleared into it, both on PullbackBuy_VWAP, and both were rejected for different reasons. 9988 BABA (Buy_Rank 1, cv64) is blocked hard by the quality gate on a zero-position new entry (Quality_Pass False, F-Score 3), with adjusted conviction 61, R:R 0.91 and a close below the 200-day average stacked behind it. 9618 JD (Buy_Rank 2, cv62) is the closest thing to a real buy this book has had - quality passes, the forward view is positive on both horizons, it trades above the 200-day average and the entry is on strength rather than a chase - but it sits 3 points below the conviction floor, prints R:R 0.92 (the HT-3 config artifact), and reports Q2 after tomorrow's HK close, so L6 forbids queueing into it. The sell pool is empty: the only holding is 0992 Lenovo and all three sell-origination classes (signal, ADR-0020 risk-discipline, amplified valuation) are checked and closed - no Exit_Warning code, not PROFIT_ARMED, no stop break with 14.5% of headroom, a +42.5% 12-week view, and a HOLD label whose conviction ROSE. No anticipate slot - no name has printed BUY - ANTICIPATE in any of the last four snapshots. The governing fact of the day is that Lenovo Q1 FY2026/27 results land TOMORROW, 2026-08-13, two-sided +/-6.0%, on the book's only position; the catalyst-aware rule is hold and trail through it, and L6 blocks adds ahead of it - which is also why the two rejected candidates would be revisited after their own 08-13 and 08-20 prints rather than before. Book value HK (0992 at a blended 27.498, MV 290,400 = 14.41%; cash 1,725,020 = 85.59%) = +0.77% vs the 2,000,000 seed, the best mark since inception. Buys need HKsells free HKall retained. Scorecard refreshed: 80 entries, 10 graded, calib_1w -0.346 still negative so f=0.30 stands (L5); every cell remains below the n>=5 floor so scorecard_bias is 0 everywhere (HT-2); BreakoutUp_20D|buy was REGRADED down to hit_1w 0.25 at n=4 with no new sample, a second consecutive session of evidence leaning L1's way. Structural note carried to the learnings review: HT-3 is now measured at 0.91-0.92 on four independent names and is silencing this book's only workable buy family - a config item gated behind champion/challenger, never to be hand-edited. |
Outcome & track record
Accuracy at the time of this record.
10 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 →
-0.346 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.