Recommendations
2 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 | 8.99 | BUY — QUEUED @8.83 | 8.83 | 8.34 | 8.99 (+1.8%) | 9.02 (+2.2%) | 8.95 (+1.4%) | China Construction Bank has broken out of a four-week base on the strongest signal this book has ever recorded, and we are buying it -- but only on a pullback, never at today's price. The shares closed at 8.99, right at the day's high and within 1% of their 52-week high, on very heavy volume, with the trend structure fully aligned above every major moving average. Conviction has climbed for four straight sessions (58, 74, 79, now 90) as the setup matured from anticipation into a confirmed breakout. What holds us back from paying up is that the model's own twelve-week projection (8.95) sits fractionally BELOW today's close -- it does not expect much more from here at this price. The resolution is simple: we queue a limit order at 8.83, the model's own entry level and 1.8% below market. At that price the twelve-week view is positive again, the four-week view is +2.2%, and a hard stop at 8.34 caps the loss at under 1% of the book. Chasing a bank into a 35%-odds Fed decision two days away is how this book has lost money before; a resting limit that only fills on weakness is how it should be staging into one. If the shares never come back to 8.83, the order simply expires and costs nothing. OPERATIONAL: STRONG BUY BreakoutUp_20D cv90, Buy_Rank 1, 90.4, Trend pillar 89. adj_conv 90 = 90 + val_adj 0 (NA_CCY, quality gate OFF) + bias 0 (BreakoutUp_20D|buy n=2, report-only). Scan Order_Type = Buy Stop @8.83 but Entry_Distance_ATR -0.83 (already through the trigger, close at day high) -> CONVERTED to a pullback GTC LIMIT at the same printed level per the no-chase rule and HF-3's queue-at-pullback branch. HF-3: Target_12Week 8.95 < Close 8.99 (-0.44%, improved from -2.4% on 07-24) = inverted vs CLOSE, but 8.95 is +1.4% ABOVE the 8.83 limit -> the queue branch resolves it. HF-4 PASSES for the first time: Est_1W 8.99 >= entry 8.83 (was 8.75 < 8.76 on 07-24 -- this is precisely the re-arm trigger the 07-24 card recorded). L1: BreakoutUp_20D is a discounted chase family, which is why this is a limit and not a market order. Sizing: 90.4% x HKx 20% buy_pct = HKnotional, modulator x1.0, /8.83 = 40,868 -> (1,000-share HK board lot) = HK = 17.7% of book; max_position_pct 60% not binding; 1,645,623.50 not binding. Stop 8.34 = -5.55% / 2.54 ATR -> risk HK (0.98% of book). Exit strategy chandelier_trail. Est_* are modelled off the 8.99 close and reprice down on an 8.83 fill. Horizon: FOMC 07-29 (~35% hike odds; a hike widens HK deposit spreads = constructive for the banks), Politburo ~07-30, China NBS PMI 07-31; CCB interim is ~28/29 Aug, outside the window. Ex-div HKwas 07-02 (spent). L8: executes at the 07-28 open as a resting limit; expires 07-30 if unfilled -> reconcile with a status:expired line. · news |
| 3328 | Bank of Communications | 7.42 | BUY — QUEUED @6.91 | 6.91 | 6.77 | 7.38 (+6.8%) | 7.33 (+6.1%) | 7.26 (+5.1%) | Bank of Communications is trending and has run 7.4% above its volume-weighted average price, so we are placing a patient limit order to buy it back at that average rather than paying the day's high. This is the first time this book has seen a mean-reversion pullback setup, and it is statistically the most reliable buy pattern in our records -- roughly six in seven have worked elsewhere, versus one in four for the breakout chases that have repeatedly stopped us out in this sector. The order rests at 6.91, nearly 7% below the 7.42 close, with a very tight 6.77 stop just underneath. If it fills, the model's twelve-week view sits 5% higher and the four-week view 6% higher, against 2% of downside -- roughly a five-to-one payoff and the best-shaped opportunity in the book. If the shares simply keep rising, the order expires unfilled and costs us nothing, which is an acceptable outcome for a patient limit. Nothing about this position requires the shares to do anything before Wednesday's Fed decision. OPERATIONAL: BUY PullbackBuy_VWAP cv72, Buy_Rank 2, 26.5, regime Trending, pillars Trend 75 / Momentum 50 / Volatility 45, RSI 64.5. adj_conv 72 = 72 + val_adj 0 (NA_CCY, quality gate OFF) + bias 0 (no PullbackBuy_VWAP cell in this book yet -- first print). Order_Type is natively Buy Limit @6.91 = exactly the VWAP 6.9073; Entry_Distance_ATR -3.60 (price 3.6 ATR above the limit, close at day high) -> queue-only is mandatory anyway. HF-3: Target_12Week 7.26 < Close 7.42 = inverted vs CLOSE, but 7.26 is +5.1% ABOVE the 6.91 limit -> queue-at-pullback branch resolves it. HF-4 PASSES: Est_1W 7.38 >= entry 6.91. L1 explicitly favours this family. Sizing: 26.5% x HKx 20% buy_pct = HKnotional, modulator x1.0, /6.91 = 15,309 -> (1,000-share board lot) = HK = 5.2% of book. Stop 6.77 = -2.03% / 0.99 ATR -> risk HK (0.11% of book). Exit strategy fixed_then_trail. Est_* are modelled off the 7.42 close and would reprice down materially on a 6.91 fill -- the quoted +5-7% upsides exist only because the limit sits 6.9% under market. Horizon: no hkfin-specific file and hk_horizon.json does not cover 3328 -> this name is horizon-blind; sector macro only (FOMC 07-29, China NBS PMI 07-31, Caixin PMI 08-03). L8: rests from the 07-28 open; expires 07-30 if unfilled -> reconcile with a status:expired line. · news |
| 3988 | Bank of China | 5.38 | BUY-STOP | 5.36 | 5.02 | 5.38 (+0.4%) | 5.43 (+1.3%) | 5.38 (+0.4%) | Bank of China qualified for our anticipation slot for the first time -- a second consecutive anticipation print at high conviction -- and we still passed. The reason is the payoff, not the pattern: the model's twelve-week estimate (5.38) is identical to today's close, meaning it expects essentially nothing over three months, while the stop sits 6.3% below the entry. Risking six to make half a percent is not a trade worth making, and it would be a buy-stop order that fills at Tuesday's open, one day before a Fed decision with live hike risk -- buying into the event rather than staging around it. The two orders we did place are limits well below market, which fill on weakness instead; this one would do the opposite. Bank of China stays on the watchlist and gets bought when the forward view actually pays for the risk. OPERATIONAL: BUY - ANTICIPATE AnticipationUp_VCP cv76, Buy Stop @5.36, close 5.38, Entry_Distance_ATR -0.18, RSI 62.6, 68.24. BOTH structural gates clear for the first time: >=2/3 ANTICIPATE prints (07-24 cv78 + 07-27 cv76) and cv>=55 (anticipate_slot_conviction_floor); HF-4 Est_1W 5.38 >= entry 5.36; HF-3 Target_12Week 5.38 NOT below close 5.38. WITHHELD on two grounds: (1) HF-3 passes only by EXACT EQUALITY (5.38 = 5.38) -- the weakest possible reading -- producing a +0.4% 12wk expected return vs a -6.3% stop (5.02) = ~0.06:1 reward-to-risk on the estimation layer; HF-4's principle (estimation layer contradicting the conviction layer -> demote, never buy on conviction alone) applies in substance even though its literal Est_1W<Entry test passes. (2) It is a buy-STOP already through its trigger, so it fills at the 07-28 open, one session BEFORE the 07-29 FOMC (~35% hike odds, +-3.5% band) -- the overlay's 'stage adds between the rate events, do not front-run them' line, on the card with the least forward edge; the two ACQUIRE limits sit below market and benefit from event weakness instead. Logged direction:none / optional:true so the counterfactual grader prices what withholding cost or saved. RE-ARM: a post-FOMC settled close with Est_12W >= +1.5% above entry (not equality) while still printing ANTICIPATE/BUY at cv>=55 -> take the anticipate slot at 0.5x size (~HKnotional,); OR a pullback toward EMA20 5.17 with the same forward view. Ex-div HKwas 07-02 (spent). Name is horizon-blind (not covered by hk_horizon.json). · news |
| 2888 | STANCHART | 226.00 | HOLD | — | 214.98 | 226.49 | 227.57 | 227.60 | We are holding Standard Chartered into Wednesday's half-year results rather than trimming ahead of them, and the case for holding got stronger this session. Conviction jumped 23 points to its best reading since we bought, the loss narrowed to 1.15%, and the shares now sit a comfortable 2 ATR above our stop -- the widest cushion of the holding period -- and far above every long-term moving average. The scan again flagged the position as ready for profit-taking, and for the fourth session running that flag is false: it triggers because the twelve-week target sits BELOW our purchase price, so the position reads as '99% of the way to target' while actually losing money. Booking a 'profit' on a losing position is capitulation with better branding. With results due in two days and a Fed decision the same morning -- where a rate hike would actually widen the bank's deposit margins -- the discipline is to hold through the event and let the stop do its job. The wealth-management strategy is on track and the last quarter was a record. OPERATIONAL: HOLD FailedBreakdown_Up_20D cv73 (up +23 from 50 on 07-24; conviction RISING -> fire-tick (b) does not fire), RSI 61.0, Buy_Rank/Sell_Rank blank, regime Ranging. Exit_Warning=PROFIT_ARMED SUPPRESSED for a 4th session per HF-2 (widened to the cost-inverted case): T12_Progress 99.3 arms only because Target_12Week 227.60 < Avg_Cost 228.63 while Gain_ATR -0.48 (a loss); fire-tick (c) (227.60 < 1.02 x 226.00 = 230.52) is the SAME inversion counted twice -> suppressed with it. No other Exit_Warning (no tick a); prints HOLD not NO TRADE (no tick d). No stop break: close 226.00 > 214.98 = 2.02 ATR cushion (widest of the holding period); close >> EMA200 186.11; L2 stop-break-AND-neg-12wk fires on neither leg; overlay override (0005/2888 hold-above-stop) endorses. CATALYST-AWARE HOLD (CLAUDE.md S9 trim discipline): StanChart H1 results Wed 2026-07-29 = 2 trading days out, INSIDE the <=5-session window, band +-5.0% (+-1.83 ATR), horizon stance hold-through; stacked with FOMC the same session (~35% hike odds; a hike widens HK deposit spreads = constructive for 2888). Do NOT pre-trim into the catalyst -- trim only on a confirmed lower-high or a decisive stop break. STOP 214.98 UNCHANGED (scan prints 215.07, a trivial +0.09 raise; held at the recorded level for continuity -- do not loosen). NO trail-raise: the armed-but-not-fired rule presupposes a profit to protect, and tightening into a -1.15% drawdown ahead of a +-5% print manufactures the whipsaw the optional_trim counterfactual (n=25, +2.44% median saved) warns against. Remaining risk from cost = (228.63-214.98) x 1,550 = HK (~1.06% of book). Model does not project the position back above cost on its own (1/4/12wk all -0.5% to -0.9% vs the 228.63 cost) -- the 07-29 print is what decides it. Reassess on the 07-30 settled close (L4). · news |
Outcome & track record
Accuracy at the time of this record.
3 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.
22 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.