Recommendations
0 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 |
|---|---|---|---|---|---|---|---|---|---|
| 1347 | Hua Hong Grace Semiconductor | 128.90 | SELL 85.46% | 128.90 | 122.67 | 134.28 (+4.2%) | 141.19 (+9.5%) | 132.20 (+2.6%) | Hua Hong reversed hard on the day, opening at 140.10, trading up to 144.50 and closing at 128.90 — a 7.9% outside-day loss that put it back under its 10-, 20-, 50- and 100-day moving averages, with only the 200-day still beneath it. The signal engine flipped to a STRONG SELL at the settled close, ranking it the second-strongest sell in the book and calling for an 85% reduction, and the wider backdrop confirms it: global chip stocks are more than 20% off their highs on worries about how AI infrastructure is being financed, an easing memory shortage as capacity expands, and fast-rising Chinese competition. The valuation screen shows the shares still priced well beyond any defensible intrinsic anchor. We are cutting the large majority of the position and keeping a small residual, accepting that the model's own forward estimates are still mildly positive — this is a tactical de-risk on a deteriorating trend, not a call that the story is over. OPERATIONAL: fresh signal-based STRONG SELL - REDUCE at the settled close (Sell_Rank 2, engine 85.46); new origination that SUPERSEDES the unexecuted 08-17 ADR-0012 profit-take card (30%) rather than re-issuing it - same reading applied to 2513 on 08-05, so HK-11 does not bind. Adjusted 58 = 55 + 0 (G2 suppresses the valuation amplify: DCF_per_share blank, Quality_Pass False, so MoS -21.77 does NOT bump the tier) + 3 (RallySell_BearTrend|sell, n=17, hit_1w.667, ACTIONABLE). Intraday trail NO TRADE 0 -> AVOID 63 -> AVOID 63 -> AVOID 55 -> STRONG SELL - REDUCE 55, i.e. the close is the most bearish print of the day (L4 authoritative). HK-9: hk sell cells grade at 1wk (.667) not 4wk (.250) - tactical de-risk only. Est 1wk 134.28 [115.51-160.01] +4.2%, 4wk 141.19 [105.22-176.46] +9.5%, 12wk 132.20 [91.09-220.23] +2.6% - all positive, the acknowledged cost. Sizing 85.46% x 3000 = (board lot 100). Remaining hold the 122.67 trail (0.45xATR, marginally tighter than the HK-12 band floor; model Stop_Price 101.20 at 2.0xATR is inside the band and NOT degenerate today). Position -19.44% vs 160.00 cost. News dated to the sector move, not to the 18-Aug session - no date-stamped article was retrievable (Yahoo 403). · news |
| 2513 | Z.AI / Knowledge Atlas | 1043.00 | SELL 30% | 1043.00 | 875.71 | 1083.37 (+3.9%) | 1023.01 (-1.9%) | 1185.38 (+13.7%) | Z.AI fell 13.3% on the session, opening at 1,211, trading as low as 990 and closing at 1,043 — decisively through the 1,126.38 protective stop we set only one session earlier, by 7.4%. Breaking a risk line that recent is the event our sell discipline exists for, so we are taking the minimum 30% off the table and keeping. The reason we are not cutting harder is the calendar: the Hang Seng quarterly index review is announced on 21 August with Z.AI a named inclusion contender, the company's first-ever interim results are due on or before 31 August, and its addition to the MSCI ACWI and Emerging Markets indices takes effect at the 31 August close, which brings mechanical index buying. We take the risk reduction the stop demands while staying invested for those catalysts. OPERATIONAL: exit basis is the DECISIVE STOP BREAK, not the label - the row prints AVOID cv42 at the settled close after STRONG SELL - REDUCE cv55 at 14:00/15:00/15:30 (L4: the close governs, so no fresh SELL-REDUCE leg). NOT a re-issue of the 07-31 retirement: that WATCH re-arm is a settled close < 862.00 and is untouched (1,043 is 21% above); this card rests on the newer 1,126.38 trail raised 08-17. A stop-break origination is not gated by the 65 conviction floor. Adjusted 45 = 42 + 0 (Val_Signal NA - valuation blind, MoS and DCF both absent) + 3 (RallySell_BearTrend|sell ACTIONABLE). No ADR-0012 arm: Exit_Warning blank, T12_Progress 87.99 < 90, Gain_ATR -3.53. Catalyst-aware hold permits a trim only on a confirmed lower-high or a decisive stop break - the break is what fired, so minimum tier 30%. Est 1wk 1,083.37 [954.03-1,194.44] +3.9%, 4wk 1,023.01 [866.58-1,329.08] -1.9% (the only negative 4wk in the held book), 12wk 1,185.38 [954.39-1,422.21] +13.7%. TRAIL RE-BASED 1,126.38 -> 875.71 on the remaining : the old line ended the session 8.0% ABOVE the market and could never bind again; 875.71 is the 2.0xATR chandelier off the 1,211 high, 1.00xATR below the close, inside the HK-12 band. HK-12: model Stop_Price 1,294.82 is 24% ABOVE the close - degenerate, rejected (6th consecutive session). Position -36.21% vs 1,635.00 cost. News NONE FOUND for 17-18 Aug (Yahoo per-ticker 403); standing reads unchanged - MSCI EM addition vs the US$4bn share-sale dilution overhang. |
| 0005 | HSBC Holdings | 162.10 | HOLD | — | 159.50 | 162.05 | 161.67 | 164.32 | HSBC eased 0.9% to 162.10 and remains the largest holding in the book at roughly HK. The model's profit-taking machinery is signalling on it — the shares have essentially reached their twelve-week price objective and the signal's confidence dropped sharply on the day — but we retired the sell case on 6 August with a clear trigger of a close below 159.50, and at 162.10 that has not happened. We hold the full position and let the 159.50 line do the work; it is now only 1.6% below the price, the narrowest margin of any holding, so the decision resolves itself quickly either way. OPERATIONAL: ARMED (T12_Progress 98.65) with TWO ADR-0012 ticks now firing - (c) 12wk 164.32 < 1.02 x 162.10 = 165.34, and (b) conviction collapsed, -26 points, the sharpest drop in the book. Two ticks would be profit_take_2 = 50% of HKby far the largest trim this book has ever declined. NOT TAKEN: retired 08-06 to a WATCH line with re-arm 'close below 159.50', untriggered. Critically, the mechanism firing is the SAME ADR-0012 profit-take the thread was retired under - it is not a new origination the way 1347's fresh signal label and 2513's broken stop are, so HK-11 binds and the distinction is applied consistently across all three names this session. Stop held 159.50 (tighter than the 158.21 chandelier). Valuation NA_CCY - MoS -10.72 is a currency-mismatch artifact (HKD price vs USD financials), not a real read; DCF_per_share 35.27 against a 162.10 close confirms the anchor is unusable. Est 1wk 162.05 [157.07-167.81], 4wk 161.67 [149.54-173.28], 12wk 164.32 [137.87-190.52]. Catalysts: FOMC minutes 19-Aug, US flash PMIs 21-Aug, Jackson Hole 28-Aug. |
| 0522 | ASMPT | 174.40 | HOLD | — | 163.69 | 178.89 | 188.81 | 195.73 | ASMPT fell 4.2% to 174.40 in the broad chip-equipment selloff and slipped below its 50-day average for the first time in weeks, but it remains the one holding in the book with a clearly positive forward view — the model projects roughly 8% higher over four weeks and 12% over twelve. It is also the only holding not flagged by the profit-taking machinery, because it is still 11% below our cost rather than sitting on a gain. We hold all with the stop at 163.69 and treat NVIDIA's results on 26 August as the next real test, since semiconductor capital-equipment demand reads directly off that. OPERATIONAL: the ONE held name with NO ADR-0012 arm at all (T12_Progress 89.10 < 90, Gain_ATR -2.25) and the BEST forward view in the book (4wk 188.81 +8.3%, 12wk 195.73 +12.2%). NO TRADE cv0 on all five intraday prints; NoTrade_Reason verbatim: 'Consensus conflict + low conviction'. Session open 179.10, high 182.40, low 173.00, close 174.40 = -4.23%; still above EMA10 170.60 / EMA20 168.84 but BELOW EMA50 171.24 - first EMA50 loss in weeks, flagged. Position -11.18% vs 196.35 cost. Trail HELD at 163.69 (tighter than the 162.85 chandelier), 1.10xATR below the close, inside the HK-12 band. Valuation GROWTH_PRICED_IN MoS -4.71 with DCF_per_share BLANK - G2 makes this anchor unreliable, so it neither originates nor amplifies anything. NVIDIA Q2 FY2027 26-Aug (HK reaction 27-Aug) and the SK 1-20 Aug semiconductor export print 21-Aug read through. |
Outcome & track record
Accuracy at the time of this record.
174 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 →
-0.249 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.