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 | 30.80 | HOLD | 30.80 | 29.40 | 31.98 (+3.8%) | 34.73 (+12.8%) | 43.89 (+42.5%) | Hold the full. Lenovo is the book's one position and it is improving on every axis: conviction rose, the momentum pillar, and the close recovered 3.4% to 30.80 on 156m shares - double the prior session's volume, and it came off a 28.80 low, which is an accumulation signature rather than distribution. The unrealised gain has turned back up from +8.3% to +12.0% (+HK) and the model's twelve-week view is +42.5%, against a DCF anchor of 45.65 that the shares still trade 32.5% below. Fundamentally the 13 Aug print is intact - AI-server pipeline US0bn (+157% QoQ), ISG revenue near doubled at a 9.1% operating margin - and the Event Horizon shows nothing scheduled against the name for the next ten sessions, so it has a clear runway. The one new negative, Netlist's 14 Aug ITC complaint over DDR5 patents, is eleven days old, already in the price, and is not the dilution overhang that would override the hold. There is no exit case: no Exit_Warning, profit-take unarmed (Gain_ATR 1.68 < 3.0), no stop break, no trail erosion. OPERATIONAL: trail HELD at 29.40 for a third session - NOT lowered to the 28.80 the row prints (a trail is never lowered; 28.80 is simply today's low, the recurring Systemic #5 artefact) and NOT raised to the chandelier's 30.26, which is executable for the first time in three sessions but sits only 0.27 ATR under the market on a day whose own range was 1.05 ATR and whose low was 28.80 - a stop today's session would have taken out is an exit trigger, not a risk line (Systemic #1). Headroom 4.55%, protecting HKat 6.92% above the 27.498 cost. The trail was penetrated intraday for the second time (low 28.80, 2.04% below) and recovered to close 4.76% above it; the trigger is a DECISIVE CLOSE below 29.40 (L4). L2 does not apply - close above stop AND 12wk positive, both legs fail. ADR-0020 not triggered (live stop = peak stop 29.40, 0% erosion). G1/G3 would block any valuation trim anyway (Trend 77, Regime Trending, T12 +42.5%). NO ADD: the 29.40-31.50 base-build zone and RSI < 65 legs are both met (30.80, RSI 60.72) but the label is HOLD/[WATCH], not actionable - the row's Entry_Price 30.80 is a reference, not a trigger (provenance rule). 1wk at f=0.30 per L5 (calib_1w +0.129 on a stale n=10 sample). T12_Progress_Pct 70.18 frozen for a 9th snapshot; hand-computed 20.1% (Systemic #2). · news |
| 6869 | YOFC | 162.50 | WATCH | 157.78 | 110.91 | n/a | 251.53 (+59.4%) | 97.44 (-38.2%) | Declined despite the strongest signal this book has produced in months. YOFC printed STRONG BUY at conviction 86 (rank 1) after closing +16.1% at 162.50, but we are not paying for it. The company fails our accounting-quality screen outright - no reliable cash-flow anchor, and the shares change hands at roughly thirteen times their earnings-power value - so this is a new position we are not permitted to open. The model's own twelve-week view is 97.44, forty per cent BELOW today's close, flatly contradicting the four-week target of 251.53 on the very same row. The move itself is the problem: the stock opened at 137.70 and closed at 162.50, up 18% intraday, finishing in the top 7% of its range and 31.7% above VWAP, with no company announcement anywhere to explain it - no results, no order win, no guidance change. That is sector flow into NVIDIA's 26 Aug print, not a company event, and this book has just watched two identical setups round-trip: the two biggest gainers of 21 Aug closed today at -13.7% and -18.2%. The risk line is also unusable - a stop at 110.91 is 31.7% below the market, and NVIDIA (26 Aug) plus YOFC's own interim-results deadline (31 Aug) are two binary events inside four sessions that such a stop cannot underwrite. Re-arm on a pullback to 140-150 that holds EMA100 on the retest, AFTER the 31 Aug interim, with a positive twelve-week view and a stop inside 1.5 ATR. OPERATIONAL: Quality_Pass FALSE (F-Score 6, no DCF, EPV 12.52) = hard block on new entries. adj cv 76 (86 + val_adj -10 at the clamp floor + bias 0) clears the 65 floor - the fusion floor is NOT what stops this. HT-4 forward-view leg fails (T12 -40.04%). Order_Type is a Buy Stop @157.78 sitting 2.99% BELOW the close (Entry_Distance_ATR -0.31) - a buy-stop below the market is a market order in disguise, so there is no queueable order here and the provenance rule forbids republishing 157.78 (Systemic #5). Stop 110.91 = 3.39 ATR / -31.75%; R:R from the executable 162.50 is 1.73. L1: BreakoutUp_20D|buy is this book's worst cell (n=4, hit_1w 0.25, hit_4w 0.00, report-only, bias 0 per HT-2). Valuation is marked to the 08-19 close of 131.30, so true MoS today is WORSE than -9.4876. Counterfactual size for the grader: 74.69% x 2,033,020 x 20% = HKx0.5 modifier = HK / 162.50 = -> (lot 500) = HK4.00% of book, risk to stop HK = 1.27% of book. Yahoo 6869.HK pages returned HTTP 503; read rests on search-index coverage, not page-verified. · news |
| 1211 | BYD | 93.60 | WATCH | 93.60 | 90.80 | n/a | 99.24 (+6.0%) | 95.86 (+2.4%) | Declined for a second time, but the case genuinely improved and it is worth saying so. BYD printed BUY at conviction 69 for the second reading running, and three of the four objections raised on 21 Aug have been answered: the twelve-week view has turned positive (95.86, +2.4%) alongside a +6.0% four-week target, and the stop has widened from a noise-level 0.47 ATR to a defensible 0.99 ATR at 90.80, which gives an honest reward-to-risk of 2.0. What still blocks it is the balance sheet, not the chart: BYD fails our accounting-quality screen with one of the weakest scores in the book and no reliable cash-flow anchor, and the shares screen expensive on earnings multiples, which docks the position enough to leave it three points short of our conviction bar. On top of that, interim results are dated 29 Aug and first trade on 31 Aug - four sessions after any entry - and a 3% stop cannot sit through a result that historically moves this name 7%. We would rather see the print first. Re-arm on a third consecutive actionable BUY AFTER 31 Aug, with the close reclaiming EMA200 at 96.31 and either the quality screen passing on the next valuation refresh or raw conviction reaching 72+ so it clears the floor through the valuation haircut. OPERATIONAL: Quality_Pass FALSE (F-Score 4, no DCF, EPV 38.07) = hard block on new entries; adj cv 62 = 69 + val_adj -7 (MoS -1.3481) + bias 0, three points sub-floor - two independent kills. HT-4 forward view now PASSES both legs for the first time. Entry_Distance_ATR 0.0, VWAP_Distance +1.60% - not extended, closed mid-range 90.80-94.85. Still below EMA200 (96.31, -2.83%), above EMA100 (91.80). Two BUY prints across three sessions, NOT a confirmed streak - no 08-24 snapshot exists (Systemic #7). Horizon stance on the 08-31 event is queue-below, band +/-7%, confidence med. FailedBreakdown_Up_20D|buy cell n=2, hit_1w 0.50, stop-first 0.50, report-only, bias 0 (HT-2/L1). Counterfactual size: 28.39% x 2,033,020 x 20% = HKx0.75 = HK / 93.60 = -> (lot 500) = HK2.30% of book, risk HK = 0.07% of book. · news |
| 1810 | XIAOMI CORP-W | 27.76 | WATCH | 27.76 | 27.41 | n/a | 30.12 (+8.5%) | 27.26 (-1.8%) | Yesterday's headline candidate is off the table and the decision to pass has already paid. Xiaomi was the closest call this book had seen in weeks - BUY at conviction 78, rank 1, clearing every mechanical test - and it was declined because it was a chase of a post-earnings move on a bad quarter: adjusted profit down 43%, revenue down 6.1%, with management calling conditions challenging. Two sessions later the shares are down 4.34% to 27.76, the signal has collapsed to WATCH at conviction 45, and the stock has lost its EMA100. The twelve-week view remains negative. Nothing to buy; no re-arm while the forward view is below the close and price sits under EMA100. OPERATIONAL: L6 withhold VINDICATED, +4.34% saved (29.02 -> 27.76). Label fell out of the actionable set within one reading, confirming the 08-21 objection that it was a single unconfirmed print that had flipped family from AVOID cv72 across the missing 08-20 snapshot. Close 27.76 now BELOW EMA100 28.88. T12 27.26 = -1.8%, still failing HT-4. Quality_Pass TRUE (F-Score 6) but the label is not actionable, so the gate is not reached. Recorded as a graded withhold, not a fresh candidate. · news |
| — | hktech | — | NO ACTION | — | — | n/a | n/a | n/a | NO ACTION both sides - the seventeenth consecutive session with the HKof intact. The buy pool produced its strongest print in months and it was still not good enough: YOFC's STRONG BUY at conviction 86 fails the accounting-quality screen outright, carries a twelve-week view forty per cent below the close, and came on an unexplained +16% day that this book has just watched two identical setups give straight back (-13.7% and -18.2% today). BYD improved on three counts but is still blocked by quality and sits three points under the conviction bar with results four sessions away. On the sell side there is nothing to sell: the single holding, Lenovo, is strengthening - conviction up, momentum up, unrealised gain back to +12.0%, a +42.5% twelve-week view and a clear event calendar for ten sessions. The only decision of substance is a defensive one: Lenovo's trailing stop stays at 29.40 for a third session, neither lowered to the 28.80 the model printed nor raised to the 30.26 the trail formula computed, because that level sits inside a single day's range and would have been triggered today. OPERATIONAL: 0 ACQUIRE / 0 DISPOSE. Book HK ( 15.15%, cash 84.85%), +1.65% vs the 2,000,000 seed - a new high-water mark. Nothing pending for the 2026-08-26 open; no queue to reprice, withdraw or expire. Anchor is a settled 17:00 HKT close (L4 authoritative). COVERAGE GAP: no 2026-08-24 EOD snapshot exists for hktech and none for the parent hk book either - second consecutive missed Monday after 08-20; all deltas span two sessions and neither BUY print has an intermediate confirmation (Systemic #7, escalated). Scorecard: entries 114, graded 10, calib_1w +0.129, no actionable cells, all biases 0 (HT-2); f=0.30 retained per L5. Fair value FetchedAt 2026-08-19 = 6.0d, fresh by the 7d rule but marked to pre-rally prices - regenerate next run. Overlay 51d stale and inert. Five PROPOSED learnings blocks now pending; ACTIVE block 38 days old; Step 8a not run (batch mode). |
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 →
0.129 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.