HK Financials — End-of-Day Verdict
1 DISPOSE x 0 ACQUIRE - RE-ISSUE the 50% trim of 3968 at the 08-14 open; mandatory trail-raise on 2888 223.61 -> 228.47
Recommendations
0 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 |
|---|---|---|---|---|---|---|---|---|---|
| 3968 | China Merchants Bank | 48.38 | SELL 50% | 48.38 | 48.10 | 48.35 (-0.1%) | 48.45 (+0.1%) | 48.66 (+0.6%) | We are selling half our China Merchants Bank position -- the same order we placed yesterday, which never reached the market. For a third straight day the model can find no coherent reason to own this stock, reading the evidence as outright conflicting rather than merely weak, and it now projects the shares going essentially nowhere for the next three months. That is a poor use of capital even in a bank we rate well on fundamentals. The stock did recover today and closed back above the risk line it broke yesterday, which is why we are selling half rather than all of it, and why the other half stays on to see China's July credit figures due tomorrow. OPERATIONAL: RE-ISSUE of the unfilled 2026-08-12 card, not a new decision -- the Auto-Executor's last_run_date is 2026-08-07, so the pending market-on-open trim was silently dropped at the 08-13 open; the open did not gap outside the +-1.5% band, so the order was valid and should have filled. Re-priced from HKto the 08-13 settled close HK (NoTrade row carries no Entry_Price -- provenance is the live close, never a stale limit); range 48.16-48.60 (+-0.25 x ATR 0.899). Reprice again if the 08-14 open gaps outside 47.65-49.11. BASIS: sustained NoTrade/Conflicted (CLAUDE.md section 9 / ADR-0012 tick (d)) -- 3rd consecutive NO TRADE cv0, NoTrade_Reason 'Consensus conflict + low conviction', Pillar_Consensus Indeterminate; STRONGER than yesterday's 2nd print. The stop-breach trigger is GONE (close 48.38 vs stop 48.10, +0.58% -- reclaimed after one session), which spends the tie-break that rounded (30%, untradeable at a 500-lot) UP to. Size held at 1,000 anyway: re-sizing a pending order downward because it failed to fill for infrastructure reasons is drift, and L3 forbids re-litigating an unexecuted de-risk daily. L2 does NOT fire (both legs fail: stop reclaimed, 12wk +0.58% positive). PROFIT_ARMED SUPPRESSED for a 3rd session on the ADR-0012 arm criterion itself -- Gain_ATR -0.30, position AT A LOSS; this is a signal-degradation trim, explicitly NOT a profit-take. Counter-evidence recorded and overridden: optional_trim counterfactual n=47 median +1.98%/1wk +5.96%/4wk SAVED by not trimming (HF-6); NoTrade|sell scorecard cell 0-for-2 on 1wk (report-only, n<5); 3968 is the book's highest-quality holding (Quality_Pass TRUE, F-Score 5, MoS_FV +0.245) and valuation is barred from originating and unable to amplify (NA_CCY, val_adj 0). Proceeds HK (+HKvs yesterday's reference), realised -HK, -> HK. ADR-0020 erosion 0% -- cut while the stop is at its peak. RESIDUAL hard rules: stop stays 48.10 (chandelier 46.58, ratchet never lowered); FULL EXIT on ANY close below 48.10, the 'decisive' qualifier is spent; re-arm to hold only if conviction recovers to >=45 with a non-NO-TRADE label AND the close holds 48.10 -- today met the price leg only, so UNMET; do NOT re-add on a bounce alone. Horizon: China July aggregate financing / new yuan loans ~08-14 is a direct mainland-bank read-through 1 session out and is two-sided -- the reason half the line rides. · news |
| 2888 | STANCHART | 238.80 | HOLD | — | 228.47 | 239.93 | 237.31 | 230.61 | We are holding Standard Chartered and raising its protective stop to HK -- for the first time, level with what we paid, which makes the position effectively risk-free. This was the stock's best day since we bought it: up 2.1% to a new high for us, and the model's confidence jumped 31 points, the largest single-day gain anywhere in the book, as the pattern shifted to a bullish coiled-spring setup. A momentum warning did print, but we are not selling a stock into a new high on a single divergence flag while every other measure is strengthening -- we tighten the stop instead and let it run. The one caution is that the model likes the next week but not the next quarter, so we are holding what we have and will not add. OPERATIONAL: TRAIL-RAISE 223.61 -> 228.47 (2.0-ATR chandelier 238.80 - 2 x 5.163838, clearing the ratchet floor by HK; ratchet only, never lowered). Stop now HKbelow the 228.63 cost -- a stop-out realises about -HKon vs -HKat the prior stop. Cushion 2.00 ATR; stop-risk HK (0.56% of book). PROFIT_ARMED + RSI_DIV: read mechanically, ADR-0012 ticks (a) and (c) both fire and mandate profit_take_2 = a 50% trim on the day the name makes a new high with conviction +31. OVERRIDDEN under ACTIVE HF-2, 14th consecutive session. (i) Tick (c) is NOT independent -- the arm is the 12wk inversion (Target_12Week 230.61 < close 238.80, -3.43%, T12_Progress 103.55; Gain_ATR 1.97 vs the 3.0 threshold, so the gain leg does not arm it), and per the EOD 2026-08-10 structural finding tick (c) T12 < 1.02 x Close is mathematically IMPLIED by any T12_Progress-derived arm -- arm and tick are one input. (ii) That leaves RSI_DIV alone, and HF-2 requires an independent bearish basis (decisive close < stop, or a fresh high-conviction formal SELL): label HOLD, no Sell_Rank, conviction RISING, close 2.00 ATR above the new stop, RSI 65.49 far from extended, above EMA5/10/20/50/100/200, consensus Consolidation_Bullish. (iii) Ticks (b) and (d) negative. SCOPE NOTE for the learnings review: HF-2 is worded for 'oversold-family' signals; it has now been applied on its structural ground across FailedBreakdown_Up_20D (08-10/11/12) and AnticipationUp_VCP (today) -- widen the wording to match the practice. Skill G1 did not cover this (AnticipationUp_VCP at cv77 vs its BreakoutUp_*/MomentumContinuation_* cv>=80 bound) -- systemic finding 3. All four 08-12 pre-commitments checked: (1) cv<=26? No, 77. (2) non-PROFIT_ARMED Exit_Warning? RSI_DIV MET and overridden above, not silently dismissed. (3) two consecutive NO TRADE? No. (4) formal SELL/REDUCE cv>=65? No. Engine Stop_Price 230.19 is HIGHER than the chandelier but belongs to the fresh 240.52 AnticipationUp_VCP entry we do not hold; the chandelier ratchet is retained for consistency and 230.19 is recorded as the escalation level on a second independent tick. ADR-0020 erosion 0% (peak_stop re-based to 228.47). Forward curve up-then-down (1wk +0.47%, 4wk -0.62%, 12wk -3.43%) -- hold with a tight ratchet, do NOT add; Entry_Distance_ATR +0.33 is below the 240.52 trigger and HF-3's non-inverted-12wk gate blocks a fresh entry anyway. PRE-COMMITMENT 2026-08-14 (re-based to cv77): trim 30% if conviction <= 57, OR a Tier-2/3 Exit_Warning prints (AVWAP_BREAK / EMA10_BREAK), OR two consecutive NO TRADE, OR a fresh formal SELL/REDUCE at cv>=65; a REPEAT RSI_DIV alone stays suppressed unless conviction also falls below 65. Full exit on a decisive close below 228.47. Next trail-raise when the close clears ~239.13. Horizon: no company-specific event inside 5 sessions; FOMC minutes 08-19/20 tagged to 2888 via HIBOR/NIM. · news |
| 3988 | Bank of China | 5.28 | NO ACTION | 5.28 | 5.18 | 5.27 (-0.2%) | 5.33 (+0.9%) | 5.26 (-0.4%) | Bank of China was the model's strongest riser today -- confidence nearly doubled on the back of the US inflation print -- but it never became a buy, and we are not buying it. The reason is simple and it is the rule this book has learned the hard way in Hong Kong financials: the model's own three-month projection sits BELOW today's price. We do not pay up for a stock the model expects to be worth less in a quarter, however strong the short-term signal looks. OPERATIONAL: label is WATCH, not a buy print -- 3988 never enters the buy pool, so this is a withheld near-miss and not a rejection of an actionable card. Conviction (+29), the book's largest riser; 62 + val_adj 0 (NA_CCY) + bias 0 = adjusted 62 vs the 65 floor, short by 3. Independently blocked by HF-3: Est_12W/Target_12Week 5.26 vs close 5.285 = -0.47% INVERTED -- the non-inverted-12wk gate is a hard block on HK-financials buy-chases, whose scorecard record in this book is uniformly stop-first. Signal family FailedBreakdown_Up_20D is the L1-disfavoured one (~0.25 hit, stop-first ~60%). Re-arm trigger: a genuine BUY/STRONG BUY label AND Est_12W > close by at least +2% AND adjusted conviction >= 65 -- all three, not any one. China LPR fixing 08-20 is a NIM risk for the mainland banks and cuts against a fresh entry here. · news |
| 0388 | HKEX | 408.20 | NO ACTION | 408.20 | 403.00 | 408.09 (-0.0%) | 412.63 (+1.1%) | 416.68 (+2.1%) | HKEX remains the name we are watching most closely and we are still not buying it. The three-month view is the best of any candidate at roughly +2.1%, but the exchange screens as expensively priced against what its own business earns, it fails our fundamental quality screen, and it reports interim results in six trading sessions. Buying a quality-gate failure straight into an earnings print is not a risk-defined entry at any price. We wait for the results to clear. OPERATIONAL: label WATCH for a second session -- no buy print, not in the pool. Conviction; 57 + val_adj -6 (MoS_FV -1.111, degree-scaled) + bias 0 = adjusted 51 vs the 65 floor, short by 14. Quality_Pass FALSE (F-Score 4, OVERVALUED_VS_RI against the exchange's own residual-income anchor) is a HARD gate on a new entry under Step 6. HF-3 does NOT fire -- Target_12Week 416.68 > close 408.20, +2.08% non-inverted -- so this is the one candidate blocked on valuation and the calendar rather than on an inverted forward view, and it is the first name to reconsider once 08-19 is behind it. Signal family is the L1-disfavoured FailedBreakdown_Up_20D. Re-arm: after the 08-19 interim results, on a genuine BUY label with adjusted conviction >= 65 -- which at val_adj -6 requires raw conviction >= 71. · news |
Outcome & track record
Accuracy at the time of this record.
7 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 →
0.184 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.