Recommendations
1 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 | BUY — QUEUED @30.91 | 30.91 | 28.80 | 32.06 (+3.7%) | 34.73 (+12.4%) | 43.89 (+42.0%) | Buying Lenovo back, but only if it proves the level it broke. We sold the whole position on 19 August when it closed decisively below the HKstop we had carried for eight sessions, and that exit was right - it banked a 20.5% gain on the tranche and the stock never reclaimed the line. Tonight is the first session since that the signal engine has printed an actionable buy on it: conviction 71 on a FailedBreakdown_Up_20D setup, pillar consensus back to bullish, and a trend structure that never actually broke - the close at 30.80 still sits above the 20-day, 50-day and 200-day averages in that order. Fair value is neutral and passes the quality screen (FAIR_VALUE, margin of safety -0.63, DCF anchor HK), so valuation neither pushes nor blocks. The business case is unchanged: record quarterly revenue up 43% year on year with AI revenue up 60%, against a multiple that is stretched after a very large run. So rather than pay 30.80 at market, we place a stop-entry at HK - the exact level that stopped us out - and let the stock come and take it. If it never trades there, we never buy, and we lose nothing. || OPERATIONAL: ACQUIRE 1 of a possible 3; the book's first proposed buy after five consecutive sessions with an empty buy pool. Fusion: raw 71 + val_adj -3 (MoS_FV -0.6273 x 5 = -3.14, clamped) + bias 0 = adj 68, clears the 65 floor. Quality_Pass TRUE so the hard gate does not engage. Size modulator 1.0x (MoS -0.63 is above the -1 tier; no cut, no UNDERVALUED uplift). PRE-WRITTEN RE-ARM from the 08-19 exit card, applied verbatim, three legs: (1) bullish-family ACTIONABLE label - PASS (BUY / FailedBreakdown_Up_20D, not WATCH/HOLD); (2) settled close above 30.91 - FAIL by HK, close 30.80 = -0.36%; (3) positive 12wk that is not the 1.425x clamp rail - FAIL STRUCTURALLY, T12 43.89 / 30.80 = 1.4250, the rail to four decimals for an eighth consecutive session. RESOLUTION, both deviations recorded: leg 3 is WAIVED because it is a defective test, not a failed one - the rail carries zero forward information and can never be satisfied while clamped (Systemic #2); critically the rail is POSITIVE-signed, so note B's negative-12wk falling-knife filter - which is what kills 6869 tonight - does not bar this name. Leg 2 is CONVERTED from a pre-condition into the FILL CONDITION via the order type: a GTC buy-stop at exactly 30.91 fills only if price trades through the recorded reclaim level and cannot fill on weakness. HS-5 is explicit that the GTC is the mechanism that enforces no-chase. HONEST CAVEAT: a buy-stop fills on a TOUCH, the re-arm asked for a settled CLOSE above 30.91 - this is a relaxation and is recorded as one. Accepted because leg 1 is satisfied, the trend stack is unambiguous (Close 30.80 > EMA20 28.80 > EMA50 25.49 > EMA200 17.50, Trend pillar Bullish/78, Regime Trending, RSI 60.7 not extended), risk is defined 1.02 ATR below the trigger, and the strict alternative (wait for a settled close, then buy at the NEXT open) is exactly the deferral that cost this book HKon 08-20. Second caveat: 0992 intraday-tagged 31.12 on the 08-24 morning file and failed there, so 30.91 has already rejected once this week - reflected in the size. PROVENANCE CLEAN: anchor Entry_Price 30.80 +- 0.25 x ATR 1.96699 = 30.31-31.29; the 30.91 trigger sits inside the band AND is a RECORDED number (the Peak Stop carried for eight sessions and the level price violated on 08-19), not an invented pullback limit. SCORECARD: FailedBreakdown_Up_20D|buy is ACTIONABLE at +3 (n=6, hit_1w 0.667) but its hit_4w is 0.000 - a cell that works at one week and fails at four. Per learnings section C the +3 is held REPORT-ONLY and NOT applied (bias 0), and the L1 chase discount is applied by hand to the SIZE instead. This is the single strongest argument against the trade and it is why the position is small. SIZING: book value = HKzero existing exposure; 0.6743 x 2,269,640 x 0.20 (buy_pct_of_portfolio, hk.ini) = HK; x 1.0 modulator; / 30.91 =; 2,000-share Lenovo board lot -> = HK = 13.62% of book. max_position_pct 60% (HK) not binding; not binding. Risk at the 28.80 stop = (30.91-28.80) x 10,000 = HK = 0.93% of book. Deliberately small: 10,000 against the 44,000 the book once held, and we are paying HK / +4.4% above the HKexit - the cost of a stop that worked, not a reason to refuse the re-entry the book wrote a re-arm to permit. ESTIMATES, f=0.30 per L5 (calib_1w -0.138 degenerate): 1wk 32.06 = 30.91 + 0.30 x (34.73-30.91), +3.7% - the horizon this cell actually hits; 4wk 34.73 = Target_Price, +12.4%, TREAT WITH SUSPICION given hit_4w 0.000; 12wk 43.89 = Target_12Week, +42.0%, but = 1.4250 x Close, THE CLAMP RAIL - annotated not modified, zero forward information. 43.89 does NOT exceed the DCF anchor 45.64 so no DCF warning applies; the constraint here is the rail, not the DCF. Scan envelope Est_Low 30.7132 - Est_High 39.8577. EXPIRY per ADR-0012: anchor + 3 trading sessions = 2026-08-28; auto-withdraw earlier if conviction drops below 65. Reprice from the live Entry_Price at each EOD - do NOT let this become a zombie limit. STOP on fill: 28.80 as printed (1.02 ATR / -6.8%), and it becomes the Peak Stop watermark for ADR-0020 from the fill. NO catalyst conflict: hk_horizon.json (regenerated 08-24, window 08-25 -> 09-07) lists NO 0992 event; the NVIDIA 08-26/08-27 read-through names 7709/0981/1347/0522/2513/7747/6869/2382 and does NOT name 0992. Jackson Hole 08-28 is macro-wide and coincides with expiry. NEWS none found, no override - Yahoo's quote-news path returned 503/stale for a ninth consecutive session on this book (Systemic #7); nearest dated item is 08-05 and concerns Chinese optical TRANSCEIVER makers, not Lenovo. L8: this is a GTC resting from the 2026-08-26 open, not a market order; if it fills, the 1wk/4wk/12wk horizons and the scorecard grade from the FILL date and price, and input/hksemiconportfolio.json stays at until the executor books it. · news |
Outcome & track record
Accuracy at the time of this record.
14 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 →
-0.138 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.