Hong Kong — End-of-Day Verdict
1 DISPOSE (2513 FULL EXIT 1,000 @ ~890.50 MKT - frees HK) + 2 ACQUIRE QUEUED (0857 6,000 @9.10 adj 81; 9618 400 @107.15 adj 76 - ~HKboth conditional). 1 stop breach: 2513, exited in full. Four 07-17 queues withdrawn (0981, 7709, 0823, 1810). Strongly +HK; HK -> ~HK.
Recommendations
2 to acquire · 1 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 |
|---|---|---|---|---|---|---|---|---|---|
| 2513 | KNOWLEDGE ATLAS | 890.50 | SELL ALL | 890.50 | — | 944.34 (+6.0%) | 920.90 (+3.4%) | 975.33 (+9.5%) | Exiting the position outright. This is a broken new listing in free-fall: down 45.5% against cost in three sessions, closing near the session low after a -22% intraday reversal, with the model printing an unbroken eight-print SELL-REDUCE at the book's #1 sell rank and the stop breached for the second reset running. There is no valuation floor to lean on (no fair-value coverage) and no long-term trend reference (no EMA200 — the listing is too young), so the only anchor is the tape, and the tape is one-way. The forward model still projects a bounce to ~975, but HK's 1-week calibration has been negative across 110 graded decisions, so that projection does not carry enough weight to keep capital in a name that has halved. Operational nuance: deliberate escalation from the model's 49.87% portion to a 100% exit — the full-exit basis under CLAUDE.md section 9 is a decisive stop-loss break PLUS a stable REDUCE, and both are present; this is a signal exit, not a profit-take, so the never-100% profit-take rule does not apply. Action price deviates from the printed Sell Limit @1008.58 (+13.3% above the close — the L3 anti-pattern of a trim that never fills): go to MARKET at the open. L3/HK-5 evidence: the identical 50% card issued on 07-17 was never executed and that lapse alone cost HKas the name fell from the 1110 open to 890.50. L7: check hktech before executing — execute once. · news |
| 0857 | PETROCHINA CO LTD H SHS | 9.99 | BUY — QUEUED @9.10 | 9.10 | 8.73 | 9.99 (+9.8%) | 9.95 (+9.3%) | 10.23 (+12.4%) | The strongest fused buy case in the book right now: a fresh pullback-to-VWAP entry — the one Hong Kong buy pattern that actually grades well — on the only large-cap holding candidate that is simultaneously cheap on intrinsic value and passes the quality screen. Fair value marks it undervalued with a positive margin of safety, the balance-sheet and cash-flow screen passes outright (rare in this book, where almost every candidate fails it), and the forward model sees roughly +12% over a quarter from the entry level. That combination — cheap, quality-clean, and on the book's best-graded buy signal — is what separates it from the rest of the watchlist. Operational nuance: the close at 9.99 sits 2.68 ATR above the model's 9.10 entry, so this is deliberately NOT chased — it is a good-till-cancelled limit that fills only on a pullback, per the no-chase rule and HK-4 (the book's no-chase queue currently grades net 'saved'). Entry range 9.02-9.18 (+/-0.25 ATR). Sizing: base HK (portion 70.59% x HK x 20% buy_pct) x1.25 undervalued+quality size modifier x0.5 book-note-B conservatism = HK -> -> 6,000 at the assumed 2,000-share board lot = HKExpires 2026-07-23 (3 sessions) or auto-withdraws if conviction falls below 65. L1/HK-2 applied: PullbackBuy preferred over every chase family in the pool. L7: may also sit in a sub-book — execute once. · news |
| 9618 | JD-SW | 120.00 | BUY — QUEUED @107.15 | 107.15 | 102.52 | 119.50 (+11.5%) | 120.63 (+12.6%) | 123.69 (+15.4%) | A second pullback-to-VWAP entry on the book's best-graded buy pattern, in a large-cap e-commerce name that fair value scores as roughly at-to-slightly-below intrinsic worth and that passes the quality screen cleanly. The signal matured properly rather than being chased — three sessions of watch-only at low conviction, then a decisive step up to a full buy print at the #2 buy rank — and the forward model carries it about +15% above the entry over a quarter. Valuation adds nothing here (the margin of safety is essentially zero), so this is a technical entry validated by a clean balance sheet rather than a bargain case. Operational nuance: the close at 120.00 is 3.14 ATR above the 107.15 model entry — the most extended candidate in the pool — so it is queued as a good-till-cancelled limit roughly 10.7% below the close and will simply not fill unless the market comes back to it. Entry range 106.13-108.17 (+/-0.25 ATR). Sizing: base HKx1.0 size modifier x0.5 book-note-B conservatism = HK -> -> 400 at the assumed 50-share board lot = HKExpires 2026-07-23 or auto-withdraws below conviction 65. L1/HK-2 applied. L7: 9618 likely also sits in hktech — execute once. · news |
| 0005 | HSBC HDGS | 157.80 | HOLD | — | 153.24 | 155.30 | 158.43 | 161.11 | The strongest hold in the book — closing at a 52-week high with conviction recovering to 78 on a volatility-contraction pattern, and comfortably above every long-term moving average. Nothing here argues for action; the job is to protect an extended winner into a bank-earnings and rate-decision cluster at the end of the month. Operational nuance: PROFIT_ARMED is spurious in this book (see the systemic note) and no deterioration tick has fired, so the correct response is a trail-raise, not a trim — stop raised from 152..24. RSI 71.95 is stretched; do not add here. · news |
| 0700 | TENCENT | 477.80 | HOLD | — | 430.00 | 477.63 | 475.64 | 486.26 | Held as a core position through a +3.5% rebound session. The model went quiet (no-trade, consensus conflict) after three sessions of a constructive pullback read, which is a one-print flip on a rally rather than deterioration — not a basis to act on. The standing override keeps this a core holding unless it closes below 430, and it closed at 477.80, well clear. Operational nuance: the recorded cost of 1.00 is a placeholder and makes the reported meaningless, so no profit-take arithmetic is trustworthy on this line — the spurious PROFIT_ARMED flag is ignored. Fix the in input/hkportfolio.json. · news |
| 1347 | HUA HONG SEMICONDUCTOR | 142.90 | HOLD | — | 132.00 | 150.13 | 147.61 | 154.11 | Still the book's top exit-watch, but the exit case did not mature — it bounced 3.0% and reclaimed the reset stop, so the stop-break-plus-negative-forward-view test that would authorise a trim is not met, and the twelve-week view remains positive at roughly +7.8%. Down 10.7% against cost with the model printing no-trade for twelve consecutive snapshots: no signal to act on in either direction. Operational nuance: trail held at 132, below the 07-17 capitulation low. Hard exit trigger unchanged — a fresh breakdown-family SELL print OR the twelve-week estimate turning negative means dispose immediately. · news |
| 0522 | ASMPT | 161.20 | HOLD | — | 146.49 | 169.07 | 172.97 | 186.06 | Held into its own earnings print. The bearish rally-sell label that appeared at the 07-17 close has already lapsed back to no-trade, the stop at 146.49 is intact with the price at 161.20, and the twelve-week view is the most positive in the book at roughly +15%. Down 17.9% against cost, but there is no signal-based sell here and valuation cannot originate one. Operational nuance: catalyst-aware hold — results are estimated 3 sessions away, so do not pre-trim into the print; a confirmed lower high or a decisive stop break is the only thing that changes this before the numbers land. · news |
| 0939 | CCB - H SHS | 8.65 | HOLD | — | 8.20 | 8.61 | 8.56 | 8.37 | Holding the largest share line in the book, but flagging it as the next candidate to deteriorate. The bank closed the session on its high after a 4.3% rally, and the model responded with an avoid label on a failed-breakout pattern — a family whose sell side has a large, reliable sample in this book. The forward twelve-week estimate is mildly negative against the current price, which is the ingredient that would combine with a stop break into an exit basis. That has not happened: the price is far above the stop, so no action. Operational nuance: trail raised from 7..20 to capture the rally — that is the mandated response to a rising position with a softening forward view. If it prints an outright SELL-REDUCE it becomes a dispose candidate the same session. · news |
Outcome & track record
Accuracy at the time of this record.
110 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 →
-0.283 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.
21 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.