HK Financials — End-of-Day Verdict
0 ACQUIRE x 2 DISPOSE - FULL EXIT 3968 (, stop break + 4th NO TRADE) and profit-take of 2888 (conviction fired the written pre-commitment), both at the 08-17 open
Recommendations
0 to acquire · 2 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 | 47.96 | SELL ALL | 47.96 | 48.10⚠ | 48.00 (+0.1%) | 48.15 (+0.4%) | 48.57 (+1.3%) | Exiting China Merchants Bank in full. The shares closed at HK, below the HKstop-loss we set in advance, and the signal engine has now been unable to form any view on the name for four straight sessions ('consensus conflict + low conviction'). More importantly the chart has broken down structurally: the price now sits below its 10-, 20-, 50- AND 200-day moving averages, having been above the last two only yesterday, and trades 2.5% under its volume-weighted average price. The model's own forward view is dead flat - it projects just +1.3% over a full quarter - so there is no return here worth carrying a broken risk line for. This is a purely technical exit: on fundamentals CMB is the highest-quality business in the book (passes the quality screen, F-Score 5), and the valuation layer is explicitly barred from originating a sell. No company news explains the fall; it was a broad HK-financials give-back that today's China credit print failed to arrest, and next week's LPR fixing is a further margin headwind. Operationally: this absorbs the 1,000-sh trim decided 08-12 and re-issued 08-13 that the Auto-Executor never filled (stale since 08-07) - trim + pre-committed residual exit = the whole 2,000-sh line. The 08-13 card pre-retired the 'decisive' qualifier on the 48.10 stop precisely so this would not be re-argued. Counter-evidence noted and overruled: breach is shallow (-0.29%/-0.17 ATR), 12wk is still positive so L2 does NOT fire, NoTrade|sell is 0-for-3 on 1wk (report-only, n<5), and the optional_trim counterfactual (n=49, +1.98%/+5.96% median saved) argues restraint. PROFIT_ARMED suppressed for a 4th session - position is at a loss (Gain_ATR -0.83), so T12_Progress 98.74 is an inversion artefact, NOT a profit-take. HF-5 checked explicitly: horizon shows the 08-14 ex-div is HSBC 0005, NOT 3968, so this is a genuine break. ADR-0020: stop 48.10 = peak stop, 0% erosion - cut at the peak, the good case. Proceeds HK, realised -HK. Range 47.75-48.17; reprice if the open gaps outside 47.24-48.68. · news |
| 2888 | STANCHART | 234.20 | SELL 31.8% | 234.20 | 228.47 | 236.19 (+0.8%) | 233.09 (-0.5%) | 225.98 (-3.5%) | Banking a third of the Standard Chartered winner. The position is up 2.4% and sits near a post-inception high, but conviction collapsed 34 points in one session and the signal rotated out of a bullish volatility-contraction setup back into the weakest pattern family in our playbook, on a -1.9% close. That is a genuine deterioration in a profitable position, so we take money off the table while the trend structure is still intact - the stop holds, the price remains above every moving average, and the one-week view is still positive, which is why this is a partial trim and not an exit. The quarterly view is the real caution: the model projects -3.5% over 12 weeks even as it likes the coming week. This executes a numeric pre-commitment written into the 08-13 card - 'trim 30% if conviction <= 57' - which fired at 43 with 14 points to spare. ADR-0012 tick accounting: ARMED via PROFIT_ARMED (T12_Progress 103.64, T12-inversion-derived); tick (a) NO (yesterday's RSI_DIV cleared, RSI 65.5 to 57.6); tick (b) YES (-34 pts, fires regardless of G1); tick (c) technically yes but EXCLUDED as collinear with a T12-derived arm (5th session of this exclusion, consistent with EOD 08-10/08-13); tick (d) NO. One independent tick = profit_take_1 = 30%. HF-2's 14-session suppression is correctly LIFTED - it reserved judgement pending an independent bearish leg, and a 34-pt collapse plus bullish-family rotation OUT is exactly that. Board lot: 30% of 1,100 =, untradeable at the 50-sh lot; adjacent lots 300 (27.3%) and 350 (31.8%); 350 is nearer and there is NO stop-break tie-breaker here (stop intact, 1.13 ATR cushion), so nearest-lot settles it. No valuation amplification available (NA_CCY, val_adj 0). Counter-evidence noted and overruled: optional_trim counterfactual n=49 (+1.98%/+5.96% median saved) and the stale 0005/2888 overlay 'hold, dip overdone' posture (40 days old) both argue against - a written numeric pre-commitment abandoned the first time it costs something is not a rule. Proceeds HK, realised +HK. RESIDUAL : stop UNCHANGED at 228.47 (2.0-ATR chandelier computes 224.02, BELOW the ratchet floor - never lowered); stop sits essentially at the 228.63 cost, residual stop-risk HK (0.21% of book). PRE-COMMITMENT 2026-08-17: full exit on a decisive close below 228.47; trim a further 30% on a Tier-2/3 Exit_Warning (AVWAP_BREAK/EMA10_BREAK), OR conviction <= 23, OR two consecutive NO TRADE, OR a fresh formal SELL/REDUCE at cv>=65; re-arm to hold undisturbed if conviction recovers >= 65 with a bullish label; next trail-raise when the close clears ~238.65. Do NOT add (4wk and 12wk both negative; HF-3 gate). L4: the settled scan close 234.20 is authoritative over a cached Yahoo 236.00 quote; the HKdividend had an 08-06 ex-date, 6 sessions past and immaterial (0.09%). Range 232.93-235.47; reprice if the open gaps outside 230.69-237.71. · news |
| 3988 | Bank of China | 5.25 | WATCH | 5.25 | 5.22 | 5.26 (+0.2%) | 5.30 (+1.0%) | 5.22 (-0.6%) | Passing on Bank of China. It was the book's strongest non-held name yesterday but gave back 10 points of conviction today, and it has never printed an actual buy signal - a WATCH at 52 is not a buy, and 52 is well under our 65 conviction floor. The model also projects the shares slightly LOWER in twelve weeks than they trade today, and our own hard-won lesson in this book is not to buy HK financials whose forward view is inverted. Nothing to act on. Detail: WATCH cv52 (62 on 08-13), below the 65 floor and not an actionable label. Est_12W 5.22 vs close 5.25 = -0.57% inverted, independently blocked by HF-3's non-inverted-12wk gate. Highest non-held conviction in the book, so recorded for the counterfactual grader. Quality_Pass FALSE (F-Score 4), NA_CCY so val_adj 0. |
| 0005 | HSBC HDGS(USD0.50) HK O'SEAS | 160.80 | WATCH | 162.61 | 160.18 | 160.41 (-1.4%) | 160.48 (-1.3%) | 163.50 (+0.5%) | Passing on HSBC. It went ex-dividend today, which mechanically explains much of its 1.4% fall and its drop in conviction - so we are not reading the weakness as deterioration - but it never printed a buy signal either way, and at 43 it is far below our conviction bar. The deciding detail is that the model's own one-week estimate sits BELOW the price it would have us pay: when the forecasting layer contradicts the entry layer, that is a low-quality entry and we stand aside. Detail: WATCH cv43 (58 on 08-13), below the 65 floor and not an actionable label. HF-5 applied - horizon confirms the 08-14 ex-div (~0.48% of last close), so the decline is mechanical, not deterioration; nothing acted on either way at. HF-4 LOW_QUALITY_ENTRY signature: Est_1W 160.41 < Entry_Price 162.61. Entry_Distance_ATR +0.77 (price below the model entry). Recorded so the counterfactual grader prices the pass. NA_CCY so val_adj 0; the same 0005/2888 overlay override that argues to hold 2888 applies here and is 40 days stale. · news |
Outcome & track record
Accuracy at the time of this record.
8 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 →
0.145 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.