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 3 Sept 2026, 16:00:00 GMT+8

HK Financials — End-of-Day Verdict

EOD 2026-09-03 (hkfin)

Recommendations

0 to acquire · 2 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
3968China Merchants Bank51.90SELL 50%51.9049.5951.81 (-0.2%)52.25 (+0.7%)51.67 (-0.4%)China Merchants Bank has run 6.7% above our cost into an overbought reversal and the model has turned an outright seller: it is the top-ranked sell in the book at STRONG SELL - REDUCE conviction 72, RSI 71, with a twelve-week view (51.67) that now sits below today's price. We bank half the position into that strength and let the rest run behind a raised stop. Label upgraded overnight from AVOID cv62 to a formal STRONG SELL - REDUCE cv72 at Sell_Rank #1 (engine 84.59 = target full exit); independently PROFIT_ARMED (Gain_ATR 2.82 on real cost 48.65, +6.68%) with ADR-0012 tick (c) firing (Target_12Week 51.67 < 1.02x51.90). HF-2 inapplicable - ReversalDown_Overbought is overbought-family, not oversold, and there is an independent formal SELL basis regardless. Sized to the 50% tier ( = of 500), escalated from yesterday's 25% because the AVOID-calibrated discount is retired by the formal REDUCE label; held back from the engine's 84.59% full exit because this is the FIRST REDUCE print and ReversalDown_Overbought|sell is n=4 report-only. SUPERSEDES the unfilled 09-02 card of 500 @51.75 - does not stack. Valuation NA_CCY so val_adj 0 and no tier bump. Stop raised 49.39 -> 49.59 (chandelier 2.0xATR, matches the engine's printed long-side Stop_Price). L8: executes at the 2026-09-04 open; reprice if the open gaps >~1.5%. · news
3988Bank of China5.95SELL 30%5.955.736.01 (+1.0%)5.96 (+0.2%)5.92 (-0.5%)Bank of China is our biggest winner at +9.4% and the most overbought name in the book at RSI 74.3, and the model's bull case vanished overnight - the twelve-week view (5.92) now sits half a percent below today's price. We bank 30% of the gain and keep the majority of the position, sized down from the rulebook's 50% because this book's record on selling 'no trade' prints is poor. ADR-0012 profit-take: genuinely armed on the trustworthy leg (Gain_ATR 3.68 >= 3.0, real Avg_Cost 5.44, +9.37%), fired by TWO ticks - (b) bull-case conviction collapse BUY - ADD cv74 -> NO TRADE cv0, a -74 drop through the -20 bar, NoTrade_Reason 'Consensus conflict + low conviction'; and (c) dead forward view, Target_12Week 5.92 < 1.02x5.95. Two ticks map to profit_take_2 = 50%, CUT ONE TIER to profit_take_1 = 30% on the NoTrade|sell actionable cell (bias -10, n=5, hit_1w 0/5 - this book has sold five NoTrade rows and been wrong at one week on every one), reinforced by the optional_trim counterfactual (n=74, +1.60%/+5.58% saved by holding). The trim still fires - ADR-0012 is mandatory and a -10 cell cannot veto it - but it fires small; net effect is a straight re-issue of the unfilled 09-02 30% at the live price. G1 inapplicable (no fresh breakout at cv>=80; ticks b/c fire regardless); HF-2 does not suppress (arm comes from the Gain_ATR leg, not T12_Progress, and the signal is not oversold-family). Action price is the anchor close - the NoTrade row prints a blank Entry_Price, so this is a market-on-open sell referenced to 5.95, not an invented limit. Stop UNCHANGED at 5.73: chandelier 2.0xATR = 5.673 is below the live stop and trails raise only. L8: executes at the 2026-09-04 open. · news
0939CCB - H SHS9.56NO ACTION9.569.239.60 (+0.4%)9.76 (+2.0%)9.72 (+1.6%)Holding China Construction Bank in full. The profit-take trigger technically fires but by the thinnest possible margin, the twelve-week view is still positive, and the signal remains a hold rather than a sell - there is no case here that beats the two names we are actually trimming. Armed on T12_prog 98.41 and tick (c) fires by just 0.036 (Est_12W 9.72 vs the 9.756 bar), the thinnest trigger in the book for a second consecutive session, on a NON-inverted +1.62% twelve-week view. Conviction is decaying but the label is still HOLD on FailedBreakdown_Up_20D, a family whose BUY cell is the book's one +6 actionable read. Loses the >=3-slot competition on merit to 3968 (formal REDUCE) and 3988 (two ticks, genuine Gain_ATR arm). Stop unchanged 9.23 - chandelier 2.0xATR = 9.097 is below the live stop, trails raise only. Ledgered direction none / optional true so the grader prices what not trimming costs or saves. · news
2888STANCHART233.40NO ACTION233.40228.47232.96 (-0.2%)230.03 (-1.4%)220.70 (-5.4%)Holding Standard Chartered untouched. The position shows a 'profit-armed' flag but there is essentially no profit to take - it is up only 2.1% - and the flag is an artifact of a twelve-week target that sits below our own. Conviction is rising, not falling. HF-2 false arm: T12_Progress 105.75% is produced by an INVERTED 12-week target (220.70, below both the 233.40 close and the 228.63 cost), not by a winner nearing objective; the position is +2.09% with Gain_ATR 1.02. HF-2 conditions precisely on an oversold/bounce-family signal (FailedBreakdown_Up_20D) with an inverted 12wk, so PROFIT_ARMED and the whole fire rule are SUPPRESSED - second consecutive session. HF-6 independently holds it: no SELL label, conviction rising, no second Exit_Warning, stop intact. Stale overlay override ('hold unless below the recorded stop') holds - close 233.40 vs stop 228.47. Stop unchanged 228.47 - chandelier 2.0xATR = 224.02 would LOWER it, and trails raise only. Valuation MoS_FV -7.40 is unusable on an NA_CCY row and amplify-only in any case, so it originates nothing. · news
0005HSBC HDGS(USD0.50) HK O'SEAS163.50BUY-STOP164.42157.72162.59 (-1.1%)162.35 (-1.3%)160.07 (-2.6%)HSBC prints the strongest bullish signal in the book but we are not buying it yet. The setup has only appeared once - our rule needs it to hold for two sessions - and both of the model's own forward estimates point below the entry price we would pay, which is the pattern that has burned this book before. Conviction is building fast -> so this is the name to watch. TRIPLE-blocked: (i) the ANTICIPATE slot requires BUY - ANTICIPATE in >=2 of the last 3 snapshots and 0005 has ONE (WATCH 40 -> WATCH 62 -> ANTICIPATE 83); (ii) HF-3's inversion extension fails - Est_12W 160.07 < 163.50 close (-2.10%), and HF-3 explicitly extends the non-inverted-12wk gate to the anticipate slot; (iii) HF-4 fires - Est_1W 162.59 < Entry_Price 164.42, the LOW-QUALITY-ENTRY signature that preceded 2888's -2.2% day-one fill on 07-17. RE-ARM: a second ANTICIPATE print AND a non-inverted Est_12W. Note no_chase_queue is running [cost] at n=7 / +4.97% wh_1w, so the cost of withholding is being tracked. · news
0966China Taiping Insurance22.00NO ACTION22.3820.1321.96 (-1.9%)21.69 (-3.1%)20.88 (-6.7%)China Taiping screens as the only genuinely cheap stock in this book, but we are not buying it: it fails our balance-sheet quality screen, and the model's own twelve-week forecast sits 5% BELOW today's price - a cheap stock the model expects to fall is not a bargain. QUADRUPLE-blocked: below the 65 floor on raw conviction 64; Quality_Pass = False (F-Score 5) hard-blocks it as a NEW ENTRY; Est_12W 20.88 is 5.09% below the 22.00 close so HF-3 rejects the inversion outright; Est_1W 21.96 < Entry_Price 22.38 so HF-4 fires. val_adj would be +5 (UNDERVALUED, MoS_FV +0.786, halved for the quality failure) taking adj conviction to 69, ABOVE the floor - so it is the quality gate and HF-3, not the floor, that stop it. The four-session whipsaw AVOID 72 -> BUY 64 -> WATCH 61 -> BUY 64 is itself a stability failure. · news
1299AIA Group77.30NO ACTION77.0272.4877.28 (+0.3%)77.45 (+0.6%)79.63 (+3.4%)AIA is the cleanest buy candidate this book has produced in weeks - it is the only name whose one-week AND twelve-week forecasts both point above the entry price, clearing both of the traps that have blocked every other candidate - but its conviction score is 58 against our 65 threshold, so we wait. This is the name to promote the moment it strengthens. Adj 58 (val_adj 0 on NA_CCY, no actionable cell for ContinuationLiteUp), 7 points short of the floor. Notable because it passes BOTH book-specific entry gates: Est_12W 79.63 > 77.30 close (+3.01%, non-inverted -> HF-3 clear) and Est_1W 77.28 > Entry_Price 77.02 (-> HF-4 clear). It also broke a three-session NO TRADE cv0 run. With no_chase_queue running [cost] at n=7 / +4.97% wh_1w and queue_expiry [cost] at n=5 / +8.68% wh_4w, the counterfactuals say this book's problem is now un-taken buys, not un-taken sells. WATCH for conviction >= 65. · news

Outcome & track record

Accuracy at the time of this record.

26 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 →
3.000 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.

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.