Recommendations
0 to acquire · 0 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 |
|---|---|---|---|---|---|---|---|---|---|
| 0939 | CCB - H SHS | 9.18 | HOLD | 9.18 | 8.89 | 9.16 (-0.2%) | 9.28 (+1.1%) | 9.27 (+1.0%) | China Construction Bank is held unchanged at. The signal engine has gone silent on it - conflicting evidence and no tradeable setup - and the position is fractionally underwater at minus 0.2%, but nothing in the price action justifies selling: the shares sit above every moving average, 1.9% below their 52-week high, and comfortably clear of our stop. CCB reports first-half results after the Hong Kong close on 28 August with a dividend decision alongside, which is the event that will settle the position one way or the other. We hold through it on the stop. OPERATIONAL: PROFIT_ARMED is a pure HF-2 artifact here - the arm comes from T12_Progress 99.03 on a position with Gain_ATR -0.12, i.e. a profit-take on a LOSS (HF-6: profit-take unavailable on a loss). Tick (c) is the same T12 fact read twice; tick (b) (cv is a NoTrade labelling artifact, not a graded conviction collapse, and the scorecard cell NoTrade|sell is n=5 ACTIONABLE with hit_1w 0.0 and bias -10 - this book has sold on NoTrade five times and been wrong every graded time. Only 1 NO TRADE print, so tick (d) needs 2. No stop break (9.18 vs 8.89 = 1.74 ATR). Trail-raise yields nothing (chandelier 8.847 < 8.89; engine prints no stop on a NoTrade row). Stop stays 8.89, peak_stop 8.89. RISK ACCEPTED AND STATED: the Event Horizon puts a +/-4% band on the 28 Aug print (reaction 31 Aug) and the stop is only 3.2% below the close, so an adverse gap goes through it. NoTrade_Reason verbatim: 'Consensus conflict + low conviction'. · news |
| 2888 | STANCHART | 236.40 | HOLD | 236.40 | 228.47 | 236.24 (-0.1%) | 237.43 (+0.4%) | 226.48 (-4.2%) | Standard Chartered is the book's best holding and is left alone at. Conviction in the setup rose again, from, the shares gained 1% on the session to close at 236.40 - the best close in over a week - and the position is now up 3.4% on cost. This is a trend doing exactly what it should, and we already banked two 30% trims on it this month; a third cut would ratchet a working position to nothing on no new information. OPERATIONAL: PROFIT_ARMED + tick (c) fire mechanically for the third consecutive session and are SUPPRESSED on the HF-2 mechanism, as on 08-21 - the arm (T12_Progress 104.38 = Close/Target_12Week) and tick (c) (Target_12Week 226.48 < 1.02 x Close) are one inverted-12-week fact read twice, and Gain_ATR 1.57 is far below the 3.0 independent-winner threshold. Conviction ROSE and price rose, so there is no deterioration to act on. optional_trim counterfactual n=53, median +1.72% saved at 1wk / +5.88% at 4wk confirms restraint pays in this book. Trail-raise yields nothing: 2.0xATR chandelier is 226.48 off the close and 227.48 off today's high, both BELOW the recorded 228.47 - the engine's own printed Stop_Price 226.48 is likewise lower and is NOT adopted (never lower a stop). Stop stays 228.47, peak_stop 228.47, cushion 1.60 ATR. UK-domiciled and not in the 28 Aug mainland-bank results cluster, but it does carry Jackson Hole (28 Aug, traded 31 Aug, +/-3.5% band) and US PCE on 26 Aug. · news |
| 3968 | China Merchants Bank | 49.38 | HOLD | 49.38 | 48.10 | 49.49 (+0.2%) | 49.81 (+0.9%) | 50.00 (+1.3%) | China Merchants Bank is held unchanged at, up 1.5% on cost. The shares reached 50.30 intraday - the highest level since we opened the position - before easing to close at 49.38, and conviction slipped from. That is a pause in a working trade, not a breakdown: the stock remains above every moving average and the twelve-week view is still marginally positive. Like the rest of the book's mainland banks it sits in the 28 August results window, so we hold on the stop rather than trading around the event. OPERATIONAL: PROFIT_ARMED (T12_Progress 98.76, Gain_ATR 0.77) + tick (c) newly fires (Target_12Week 50.00 < 1.02 x 49.38 = 50.37, having cleared on 08-21 at 50.70 > 50.43). SUPPRESSED on the HF-2 mechanism - arm and tick (c) are the same T12 relation, Gain_ATR 0.77 << 3.0, no independent winner basis. Tick (b) does not fire (-11 pts, under the 20-pt bar); no second Exit_Warning code; no NO TRADE run. Trail-raise yields nothing on the book's established close-based method: 2.0xATR chandelier 47.49 (= the engine's own printed Stop_Price) < recorded 48.10. Stop stays 48.10, peak_stop 48.10, cushion 1.35 ATR. Noted for the record: a HIGH-based chandelier (50.30 - 1.893 = 48.41) would permit a small raise, but switching methodology mid-position would be an undisclosed change - flagged as a systemic finding rather than applied silently. · news |
| 0005 | HSBC HDGS(USD0.50) HK O'SEAS | 163.20 | NO ACTION | 163.20 | 162.40 | 163.30 (+0.1%) | 162.93 (-0.2%) | 162.66 (-0.3%) | HSBC prints the day's highest-ranked buy signal and we are declining it. The problem is that the model's own forward estimates point nowhere: its central four-week price of 162.93 and twelve-week price of 162.66 both sit BELOW today's close of 163.20. We would be buying a stock the model expects to be lower in three months, on a signal family that has never worked in this book. This is the same name at effectively the same price we queued on 17 August and withdrew the next day - and the setup is strictly worse now, because back then the twelve-week view was positive by 1.25% and today it is negative. A buy signal with a dead forward view is not an opportunity, it is a trap. OPERATIONAL: adjusted conviction 68 clears the 65 floor and Entry_Distance_ATR 0.00 means no chase, so the rejection is HF-3, not the floor: HF-3 forbids buying inverted/negative-12wk names in this book even at STRONG BUY cv>=69, and here BOTH the 4wk and 12wk legs are inverted. HF-4 clears only marginally (Est_1W 163.30 vs entry 163.20, +0.06%). Pillars are thin behind the Trend score of 70 - Momentum 41, Volatility 20. Valuation is uninformative (NA_CCY). RE-ARM: a BUY print at cv>=65 with Est_12W above the close. Ledgered optional/withheld so the counterfactual grader prices what the rejection cost or saved. Ex-dividend HKpassed 2026-08-13 (immaterial). Carries Jackson Hole 28 Aug (+/-3.5%) and US PCE 26 Aug per the parent-market horizon. · news |
| 2388 | BOC Hong Kong Holdings | 50.40 | NO ACTION | 50.40 | 50.05 | 50.20 (-0.4%) | 50.50 (+0.2%) | 51.73 (+2.6%) | BOC Hong Kong signals a buy for the second consecutive session and is blocked for the third. Its balance-sheet quality screen fails - a Piotroski F-Score of 4 out of 9 - and our rules do not allow opening a brand-new position in a company that fails the fundamental screen, however good the chart looks. The valuation work also puts the shares slightly above what the residual-income model says the bank is worth. The technical case is real, so this stays on the watchlist rather than the reject pile, but it needs either a better fundamental print or an existing position to be an add rather than a new entry. OPERATIONAL: adjusted conviction 67 (raw 68, val_adj -1 from MoS_FV -0.271) clears the 65 floor; blocked solely by the HARD Quality_Pass=False gate on new entries (, so the ADD exemption does not apply). HF-4 also fires: Est_1W 50.20 < Entry_Price 50.40. Volatility pillar is weak at 25 and momentum 41. Est_12W 51.73 exceeds the residual-income fair value of 40.12 - projection assumes growth beyond the priced-in level. THIRD consecutive session blocked; ledgered optional/withheld for counterfactual grading. STANDING DEFECT (unchanged from 08-21): the quality gate only bites on the two HKD/HKD names it can price (0388, 2388) while every mainland bank returns NA_CCY and passes with the gate switched OFF - it screens exactly the subset it has data for and waves through the subset it does not. Code-level finding for swingtrade/fairvalue/, not licence to override. · news |
Outcome & track record
Accuracy at the time of this record.
16 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 →
1.330 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.