This website is for educational purposes only and does not constitute investment advice. Always do your own research and assess your own risk tolerance before making investment decisions.

HK End-of-Day Verdict 21 Aug 2026, 16:00:00 GMT+8

Hong Kong — End-of-Day Verdict

1 ACQUIRE + 1 DISPOSE + 0 ANTICIPATE

Recommendations

0 to acquire · 1 to dispose

TickerNameCloseActionAction 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
0823Link REIT39.76WATCH — queued order lapsedn/an/an/aQueued buy @39.81 lapsed — closed 39.76 below the level; awaiting reprice at the next end-of-day run. Link REIT has finally broken out of seven weeks of dead money: the settled close of 39.76 clears EMA20/50/200 together, the signal engine printed STRONG BUY - ADD at conviction 86 (Buy_Rank 2, Trend pillar 82), and RSI at 62.9 means we are buying a fresh 20-day breakout rather than chasing an extended one. Fair value is neutral (GROWTH_PRICED_IN but margin of safety is essentially zero at -2.1%), so valuation neither helps nor blocks; no news in the window. Risk is tightly defined - a 2.0xATR stop at 38.70 puts HKat risk against a 42.62 swing objective, roughly 2.5:1. Buy-stop at the model entry 39.81 (entry band 39.67-39.95), = 2,800 x 70.29% rounded DOWN to the 100-share board lot, size mod 1.0x; qty_after 4,700. HK-1 chase test PASSED on all three legs (RSI 62.9 < 70, second consecutive bullish print 15:30 BUY-ADD 66 -> 17:00 SB-ADD 86, VWAP reclaimed at +2.02%). Quality_Pass False does NOT bind - the hard quality gate blocks new entries only, and this is an ADD to a held name. Scorecard haircut -2 applied (BreakoutUp_20D|buy n=10, hit_1w.40, hit_4w.125 - ACTIONABLE). CAVEAT, stated plainly: all three model horizons are FLAT (1wk +0.18%, 4wk +0.05%, 12wk +1.31%) and the row is PROFIT_ARMED at T12_Progress 98.59 - the forward model sees almost nothing left above here, so this is a risk-defined breakout trade to Target_Price 42.62, not a forward-estimate trade. Target_12Week 40.33 EXCEEDS DCF_per_share 37.84 - projection assumes growth beyond priced-in. Avg_Cost is BLANK so ADR-0012 cannot compute real profit progress and the PROFIT_ARMED flag is not actionable (no-guess rule). Rate-event risk acknowledged: Jackson Hole keynote 28-Aug and US PCE 26-Aug are two-sided for a REIT. Stop 38.70 applies to the full 4,700 post-fill. WITHDRAW if not filled by 2026-08-26 or if conviction drops below 65.
2513Z.AI / Knowledge Atlas1129.00SELL 30%1129.00975.001160.11 (+2.8%)1128.65 (-0.0%)1286.06 (+13.9%)A partial de-risk on the book's largest and most-underwater position. Z.AI printed a fresh STRONG SELL - REDUCE at conviction 76 into the settled close (Sell_Rank 2, FailedBreakout_Down_20D - the book's best-sampled sell cell at n=62), the structure is a confirmed lower-high sequence -> 1129, close below both EMA20 1166 and EMA50 1254), and at x 1,129 the position is HK = 20.2% of the book while sitting 31% below cost. Trimming 30% banks HKand cuts single-name concentration to roughly 14%, while deliberately leaving exposed to the two structural catalysts now six sessions out. Valuation is blind here (Val_Signal NA, no DCF) so no amplify or tier bump. News: none found in the 08-20/21 window; the most recent dated item is 10-Jul (share sale to fund R&D + GLM-5.2 open-sourced, stock then HK) - stale, mildly confirming as a dilution overhang. SIZE DEVIATION, stated explicitly: the engine asked for 85.46% = (800 after lot-rounding); this card takes 30% = instead. Basis is the CATALYST-AWARE HOLD clause - MSCI ACWI inclusion effective at the 31-Aug close and maiden interim results due on/before 31-Aug (LR 13.49(6), announceable any session from Monday), with the Hang Seng quarterly review announced today. That clause permits a trim only on a confirmed lower-high or a decisive stop break; the lower-high IS confirmed so the trim fires, but dumping 80%+ of the position INTO a confirmed index add is precisely the pre-trim the clause forbids. Second reason to size down: the model's own forward view contradicts a thesis exit - Est_1W 1160.11 is +2.75% ABOVE the action price and Est_12W 1286.06 is +13.9%, with Est_4W flat at -0.03%. Per HK-9 an hk sell signal is a ONE-WEEK tactical instrument (hit_1w.525, hit_4w.356) and its 4wk decay is not a thesis; this is a 1-week de-risk, not an exit. HK-11 does NOT bind: the 08-18 card was WITHDRAWN on 08-19 when the trigger vanished (AVOID cv54, Portion 0, Sell_Rank blank), so the consecutive-session chain is broken and today is a fresh origination, not a fourth re-issue. Also NOT a profit-take - Gain_ATR -3.77 and the row carries no PROFIT_ARMED flag. Market-on-open at the 1,129.00 anchor close-reference (band 1,095.48-1,162.52); reprice if Monday's open gaps >~1.5%. Residual : trail RAISED 875.71 -> 975.00, set just under the Est_1W_Low band of 990.67 (-1.15xATR, HK-12 pass - the model's own Stop_Price 1,128.99 sits AT the close and is degenerate, rejected). qty_after 700.
0700Tencent457.00HOLD457.00430.00451.80 (-1.1%)450.07 (-1.5%)457.36 (+0.1%)Prints the book's #1 sell - STRONG SELL - REDUCE cv76, Sell_Rank 1, 85.46, adjusted 77 - and is deliberately NOT acted on. The exit thread was RETIRED on 08-17 to a 430.00 WATCH line, which is also the news overlay's own standing override, and the settled close of 457.00 is 6.3% above it; HK-11 forbids re-issuing a retired exit. Trend is genuinely weak (below EMA200 499.89, Est_12W 457.36 = +0.08% dead money) but the position is = HK1.6% of the book - not worth spending the escalation. Kuaishou's 19-Aug interims were the nearest read-through and 1024 has since faded to AVOID cv52, mildly confirming the caution. RE-ARM: decisive settled close < 430.00. DATA ERROR unchanged: Avg_Cost 1.00 produces Unreal_PnL_Pct +45,600% and Gain_ATR 38.77, which falsely sets PROFIT_ARMED - systemic finding #3, needs a real in input/hkportfolio.json.
1347Hua Hong Semiconductor114.80HOLD114.80105.00120.97 (+5.4%)130.76 (+13.9%)134.33 (+17.0%)The 08-19 HK-11 retirement is vindicated one session on. After four unexecuted exit cards the thread was retired to a 105.00 WATCH line; today the sell label is GONE - AVOID cv64, Sell_Rank blank, 0.0 - and the close has recovered 113.70 -> 114.80, back above EMA200 115.87 in touching distance. Had the 08-18 card been re-issued a fifth time it would now be repriced into a bounce. All three forward estimates are positive (1wk 120.97 +5.4%, 4wk 130.76 +13.9%, 12wk 134.33 +17.0%), which HK-9 reconciles - hk sells are a one-week instrument. Position -28.25% vs 160.00, = HKRE-ARM unchanged: decisive settled close < 105.00. STAND DOWN: reclaim > 125.00. HK-12: no usable model stop on an AVOID row (Stop_Price prints AT the close, 114.80); the 105.00 line stands.
7709CSOP SK Hynix Daily (2x) Leveraged ETF39.64HOLD39.6428.0040.46 (+2.1%)38.79 (-2.1%)41.19 (+3.9%)Sixth consecutive session retired, and the settled close supports it - AVOID cv60, not a sell label, stable across all four of today's prints. The 08-14 re-arm was 'decisive close < 28.00 OR a breakdown-family signal'; the close of 39.64 is 42% above the line and today's signal family is failed_breakout, not breakdown. The leveraged-2x rule caps any future action at a 50% trim =. South Korea's 1-20 August semiconductor export line printed today and NVIDIA's Q2 lands 26-Aug - both are live two-sided catalysts for a 2x SK Hynix tracker, another reason not to force an exit here. HK-12 RECURS: model Stop_Price 51.89 sits 30.9% ABOVE the 39.64 close - an INVERTED stop on a long, rejected (systemic finding #4, a sign error worth a sanity check in signals/exits.py). Trail HELD 28.00. Position -65.40% vs 114.56; whole position HKThe overlay's standing '7709 close below 101 -> full exit' override is 47 days stale and was blown through long ago.
0522ASMPT161.70HOLD161.70155.02166.40 (+2.9%)180.94 (+11.9%)177.37 (+9.7%)Yesterday's stop breach has healed: the close recovered 158.50 -> 161.70, back 4.3% ABOVE the 155.02 trail re-based on 08-19, so L2 never gets its second leg. NO TRADE cv0 with NoTrade_Reason quoted verbatim: 'Consensus conflict + low conviction'. Forward view remains the best in the book - Est_1W 166.40 +2.9%, Est_4W 180.94 +11.9%, Est_12W 177.37 +9.7% - which is exactly why L2 requires stop-break AND a negative 12wk before trimming. PROFIT_ARMED is SPURIOUS: T12_Progress 91.17 crosses the 90 line but Gain_ATR is -3.62 and the position is -17.65% vs Avg_Cost 196.35 (systemic finding #1). Quality gate F and MoS -4.71 GROWTH_PRICED_IN neither originate nor amplify a sell on a held name (CLAUDE.md s9). NVIDIA's 26-Aug results are the nearest read-through. Trail HELD 155.02.
0939CCB - H9.24HOLD9.248.889.23 (-0.2%)9.36 (+1.2%)9.20 (-0.5%)Still the steadiest holding - HOLD on all six prints, conviction -> ->, close 9.245 within a whisker of the 52-week high 9.36, Trend pillar 76. PROFIT_ARMED (T12_Progress 100.49) and ADR-0012 tick (c) DOES fire - Est_12W 9.20 is below 1.02 x 9.245 = 9.43, a dead forward view - but NO profit-take is issued: Avg_Cost is BLANK, so real profit progress is unknowable and ADR-0012 explicitly forbids guessing when Cost is unknown. Armed-without-actionable-arm therefore resolves to the mandatory trail-raise: 8.68 -> 8.88, the model's own Stop_Price at 2.03xATR below the close (HK-12 pass). Largest position by share count, = HKInterim results + interim dividend decision land 28-Aug - a direct, dated catalyst; hold through it per the catalyst-aware rule. RE-ARM unchanged: fresh SELL-REDUCE or decisive close < 8.88.
0981SMIC72.50HOLD72.5069.6372.50 (+0.0%)75.01 (+3.5%)78.47 (+8.2%)Recovered to HOLD cv47 on all four of today's prints after two sessions of NO TRADE, which extinguishes ADR-0012 tick (d). Close 72.50 sits above EMA20 70.58 and EMA200 68.90; forward view positive at Est_4W +3.5% and Est_12W +8.2%. PROFIT_ARMED (T12_Progress 92.39) but Avg_Cost is BLANK - ADR-0012's no-guess rule blocks any profit-take, same as 0939. Trail HELD 69.63 (0.77xATR below the close; the model's own 65.05 is looser and rejected in its favour). MoS -5.31 GROWTH_PRICED_IN does not originate a sell on a held name. Two dated catalysts: Korea's semiconductor export line today and NVIDIA 26-Aug. Note the foundry pair continues to decouple - 0981 held while 1347 sits 36% below its own EMA20.
2888Standard Chartered234.00HOLD234.00229.36233.81 (-0.1%)235.66 (+0.7%)225.89 (-3.5%)HOLD cv60 at the close, close 234.00 just 2.5% off the 52-week high 240.00 and far above EMA200 194.60. PROFIT_ARMED (T12_Progress 103.59) with tick (c) firing again - Est_12W 225.89 is below 1.02 x 234.00 = 238.68 - but the thread was RETIRED on 08-13 with re-arm 'decisive close < 230.19 OR fresh SELL-REDUCE' and neither has triggered; HK-11 binds, a retired mechanism re-firing is not a new origination. Avg_Cost is BLANK in any case, so ADR-0012's no-guess rule independently blocks a profit-take. Trail HELD 229.36 (0.94xATR, tighter than the model's 224.74); the 230.19 re-arm sits just 0.83 above it so one decisive session resolves this either way. Smallest position in the book, = HKJackson Hole 28-Aug is the live rate catalyst.
0005HSBC Holdings162.90HOLD162.90159.50163.07 (+0.1%)162.83 (-0.0%)162.42 (-0.3%)Strengthened materially through the session - HOLD -> -> - and closed 162.90, up 1.05%, within 4.0% of the 52-week high 169.70 and 18.6% above EMA200 137.38. The exit thread was RETIRED 08-06 to a 159.50 WATCH line and the close is 2.1% above it, so HK-11 binds. PROFIT_ARMED (T12_Progress 100.30) but Avg_Cost is BLANK - no-guess rule, no profit-take. Forward view is flat (Est_1W +0.10%, Est_4W -0.04%, Est_12W -0.29%): dead money but not a sell, and at = HKthis is the book's single largest position by value at 34.9% of holdings. Trail HELD 159.50 (1.35xATR below the close, tighter than the model's 157.87). HK July CPI printed today, HK external trade 25-Aug, Jackson Hole 28-Aug - all rate-relevant, all two-sided.

Outcome & track record

Accuracy at the time of this record.

177 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 →
-0.261 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.

This website is for educational purposes only and does not constitute investment advice. Always do your own research and assess your own risk tolerance before making investment decisions.