Singapore — End-of-Day Verdict
28th cycle: OCBC's profit-take FIRES for real as the signal collapses cv90 → cv0 on the book's ONE genuinely-armed position (+20.9%, RSI 77.6, dead 12wk); DBS reprints STRONG BUY - ADD cv90 at BuyRank #1 and its queue is repriced 75.87 → 76.41 with the window SHORTENED to dodge the 08-14 81c ex-dividend; and B61/HMN carry into cycles 2 and 3 with the holdings file static for an 18th consecutive cycle
Recommendations
1 to acquire · 3 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 |
|---|---|---|---|---|---|---|---|---|---|
| D05 | DBS Group Holdings | 76.99 | ADD — QUEUED @76.41 | 76.41 | 73.54 | 76.87 (+0.6%) | 76.00 (-0.5%) | 79.82 (+4.5%) | Singapore's largest bank has just delivered its strongest quarter on record - total income crossed six billion dollars for the first time and profit rose nine percent - and the shares have held the resulting breakout for three sessions rather than giving it back, which is the pattern we want to see before adding. It ranks first in the book on signal strength. It is not cheap on intrinsic worth, so we are not chasing it: we place a limit order slightly below the current price and let the market come to us, and we deliberately let that order lapse before the shares go ex-dividend so we are never filled on a mechanical price adjustment rather than a real pullback. A large dividend is paid on 14 August. OPS: cv90 BuyRank#1, adj 87 (val_adj -3 MoS -0.599, bias 0 BreakoutUp_20D|buy n=4 report-only). REPRICES the unfilled 08-07 queue 75.87 -> 76.41 per the provenance rule. eD -0.40 fails the -0.25 no-chase bar and RSI 75.3 is hot, so QUEUED not market; SG-6 immediate-tranche denied (cv90 < 95, RSI-hot). Model prints 'Buy Stop @ 76.41' below a 76.99 close - executed as a GTC LIMIT instead, a buy-stop there fills instantly. Sizing: model S (114.38pct x 200k x 20pct), size_mod 1.0x, capped by declared S -> 315 -> (100-sh lot) = Sresidual S. expires_on SHORTENED to 08-13 (2 sessions not 3) to stay clear of the 08-14 81c ex-dividend (66c ordinary + 15c capital return, 1.06pct of price); if unfilled, re-queue post-ex at 75.60. Est_4W 76.00 is -0.5pct vs action price BUT that is the ex-div mark-down, not deterioration - total-return 4wk is approx +0.6pct. Est_12W 79.82 exceeds the DCF anchor 52.03 (bank valued on residual income, DCF a weak anchor here). Target_Price 85.46 is the risk objective, distinct from the 4wk estimate. L6 clears: earnings were 3 sessions ago and the gap has held, this is consolidation not a gap-chase. Trail raised 73.28 -> 73.54. · news |
| B61 | BT Sembawang Estates | 4.53 | SELL 30% | 4.53 | 4.62⚠ | 4.53 (+0.0%) | 4.52 (-0.2%) | 4.56 (+0.7%) | The shares have now closed below our risk line for a third straight session and the model has no constructive view on them at all, while the projected path over the next month is flat to slightly lower. That combination - a broken stop with nothing ahead - is our standard reason to reduce. We sell three tenths of the holding rather than more, because on intrinsic worth this is the one genuinely cheap, well-covered name we own, and selling something cheap argues for the smallest sensible cut. The remaining stake keeps its dividend, payable 15 August. OPS: L2/§9 basis - close 4.53 vs the ex-div-rebased 4.62 stop, 3rd consecutive session below (4.55 / 4.54 / 4.53), NO TRADE cv0 'Consensus conflict + low conviction', RSI decaying 39.6 -> 38.6 -> 37.5. Est_1W 0.0pct, Est_4W -0.2pct, Est_12W +0.7pct (indistinguishable from zero over a quarter). SG-1 ESCALATION at cycle 2: the 08-07 card was not executed at today's open, so this converts from a limit to an unconditional MARKET order at the open. Portion held at 30pct - valuation amplifies DOWN (UNDERVALUED, MoS +0.216, nowhere near the <=-3 that bumps a tier). NOT the 08-05 false positive: that was the 22c ex-dividend and the stop was correctly rebased to 4.62 then; tonight's break is on the rebased line. Residual, stop held 4.62, tranche 2 re-arms on a close below 4.50 or a negative Est_12W. Sdividend receivable on the pre-trim 26,300, unaffected by an 08-12 sale. · news |
| O39 | OCBC Bank | 31.36 | SELL 50% | 31.36 | 29.73 | 31.34 (-0.1%) | 30.82 (-1.7%) | 31.53 (+0.5%) | We are up almost twenty-one percent on this bank and the case for the last stretch of that gain has just come apart: the model's conviction collapsed from its highest reading to nothing and stayed there through every check of the session, momentum is diverging at a very overbought reading, and the twelve-week projection now sits essentially at today's price. The results themselves were excellent - record first-half profit, a fifteen percent dividend increase - and that is precisely the point; the market has already paid for them, and commentary is now asking whether the shares are fully valued. We bank half the position and keep half, with a tighter trailing stop under it. OPS: ADR-0012 profit-take ARMED + 3 TICKS. SG-7 sanctions this - the ONLY holding with a real Avg_Cost (25.93, human add 07-15); every other PROFIT_ARMED in this book is spurious. ARMED: Gain_ATR 6.66 >= 3.0, T12_Progress 99.46, unrealised +20.93pct. Tick (a) RSI_DIV at RSI 77.6. Tick (b) conviction, stable across all six of today's prints (not a settle flicker). Tick (c) Target_12Week 31.53 < 1.02 x 31.36 = 31.99. G1 does NOT protect: the softened rule needs a fresh BreakoutUp/MomentumContinuation at cv>=80 and today's row is NoTrade cv0 - this is the deterioration G1 was waiting for on 08-07 when it correctly converted the same fire to a trail-raise while conviction was RISING. G2/G3 govern the valuation-elevate path only and would not block it anyway. Portion profit_take_2 = 50pct (3 ticks, past the 2-tick escalation); valuation adds no bump (MoS -0.608, not <=-3); never a full exit from a profit-take alone. Third real-cost fire, following the validated 07-15 (30pct) -> 07-17 (50pct) sequence. Banked gain 500 x (31.36 - 25.93) = SResidual, stop RAISED 28.51 -> 29.73 (2.0 x ATR 0.8146 chandelier). DISCLOSED COST: ex 47c on 08-17, so an 08-12 sale forfeits Sof dividend (1.50pct of trim) against a -1.7pct 4wk path - disclosed, not decisive. Est_12W exceeds the DCF anchor 25.10 (residual-income bank method, weak DCF anchor). SG-4 check: this is NOT a trim on RSI alone - it is a fired profit-take on a real with a conviction collapse and dead 12wk alongside, and of core remain. CAVEAT: cv0 is driven by 'Liquidity gate failed' on 13.25m shares - see systemic finding 3; the verdict does not rest on tick (b) alone. · news |
| HMN | CapitaLand Ascott Trust | 0.86 | SELL ALL | 0.86 | 0.88⚠ | 0.86 (+0.0%) | 0.86 (+0.0%) | 0.87 (+1.2%) | This has closed below our risk line for a fifth straight session and now trades under every major trend average in a confirmed breakdown pattern. There is genuinely good news attached to it - a hundred-and-thirty-four-million dollar acquisition management says lifts distributions by 2.4 percent, plus addition to the index reserve list - but the price has not moved a cent on any of it in five sessions. News the market declines to price is not a reason to keep holding a broken chart; it is a reason to watch the name for a re-entry later. We sell the lot, which is a very small stake, in a single ticket. OPS: §9 signal-based exit - 5th close below the 0.88 stop, BreakdownDown_20D, below EMA20/50/200; STRONG SELL - REDUCE cv13-20 through EVERY intraday print today before settling AVOID cv48, so the bearish read was the session's constant. Full sweep not a tier: = Sis a de-minimis stub whose partial residue would cost more in commission than it protects. Est_1W/4W both 0.0pct, Est_12W +1.2pct - dead flat for a quarter. Est_12W exceeds the DCF anchor 0.735. News ASSESSED and does NOT override: constructive but wholly unconfirmed by the tape. WARNING - SG-8 THRESHOLD REACHED: this is cycle 3 (08-06 -> 08-07 -> tonight) with the monitor reporting no change. If still unexecuted at the next EOD the card will NOT be re-published; it converts to a single blocking 'holdings not synced - reconcile input/sgportfolio.json' flag. Size is already correct at 100pct and will not be re-escalated. Re-arm a watch on a reclaim of 0.88 with a bullish family print. · news |
| J69U | Frasers Centrepoint Trust | 2.23 | HOLD | 2.23 | 2.18 | 2.23 (+0.0%) | 2.23 (+0.0%) | 2.26 (+1.3%) | The model is shouting to sell this REIT - it is the single loudest sell signal in the book tonight - and the evidence underneath it does not support the call. The price is above the level we said would define a break, it rose today rather than fell, momentum improved, and the twelve-week view is positive. This specific sell pattern has been right almost exactly half the time across seventy-three graded cases in this book, which is a coin flip. We do not act on a label alone. No action. OPS: SG-2 FAILS for the 8th time. STRONG SELL - REDUCE cv71, Sell_Rank #1, portion 80.07pct - and: close 2.23 is ABOVE the recorded 2.17 stop and the live 2.18; price ROSE (2.21 -> 2.23); RSI improved 36.1 -> 43.5; Est_12W 2.26 = +1.3pct positive. SG-2 requires stop-break + negative 12wk. Live scorecard has FailedBreakout_Down_20D|sell ACTIONABLE at 0 (n=73, hit.500) - neutral across 73 grades, and SG-5 explicitly rejects promoting it. Stop RAISED 2.17 -> 2.18 to the live model line. Watch: C38U 1H results 08-12 read across to the whole retail-REIT complex. |
| U11 | United Overseas Bank | 42.16 | HOLD | 43.36 | 42.15 | 42.19 (-2.7%) | 42.24 (-2.6%) | 42.29 (-2.5%) | Our largest holding is the closest thing to a warning on the board tonight: the model's conviction in it has fallen steadily over three sessions and the price closed one cent above the level at which we would act. It is not a sale yet - the line has not broken and the forward view is still marginally positive - but it is the name to watch tomorrow. Note that the shares go ex-dividend on 17 August, which will mechanically drop the quoted price by about two percent; we are moving the risk line down by the same amount on that date so a bookkeeping adjustment is not mistaken for a real break. OPS: conviction collapse -> 26 across three sessions; close 42.16 vs stop 42.15. NOT a dispose - no stop break, Est_12W 42.29 positive (+0.3pct), eD +1.34 puts it well BELOW the model entry (a pullback, not a breakdown). MANDATORY REBASE: on 08-17 the 88c ex-dividend marks the quote down 2.09pct - stop MUST move 42.15 -> 41.27 that morning or the monitor reports a false breach on Sof stock (the B61 08-05 failure mode). A close below 42.15 BEFORE 08-17 is a genuine L2 trigger and makes U11 the lead dispose. SG-7: its PROFIT_ARMED is spurious (no Avg_Cost) - see systemic finding 1, this row plus 2 clean ADR-0012 ticks would otherwise have trimmed S on a fabricated. |
| C38U | CapitaLand Integrated Commercial Trust | 2.51 | WATCH | 2.51 | 2.38 | 2.51 (+0.0%) | 2.52 (+0.4%) | 2.51 (+0.0%) | A strong-looking breakout to a fifty-two-week high on very heavy volume, and we are not buying it. The company fails our basic financial-health screen, which blocks new positions outright; on intrinsic worth it is the most expensive name we track by a wide margin; and it reports half-year results tomorrow, so opening a position tonight would be a coin flip on an announcement. Three independent reasons, one answer. OPS: quality gate HARD BLOCK - Quality_Pass False (F-Score/FCF screen) and the name is unheld at, so the gate bites (it never blocks ADDs to held names). Would fail anyway: MoS_FV -43.12 is the book's most extreme -> val_adj -10 -> adj 63, two under the 65 floor. L1 warns BreakoutUp_52W 4wk targets overshoot +55-75pct and mean-revert. C38U 1H 2026 results 08-12 per the horizon file. BuyRank #2, 32.1m sh traded. |
| F34 | Wilmar International | 3.96 | WATCH | 3.92 | 3.82 | 3.98 (+1.5%) | 3.96 (+1.0%) | 3.92 (+0.0%) | This is the entry shape we like best - a pullback into support rather than a chase - and it passes the financial-health screen, so it was a live candidate. It falls five points short of our action threshold once we mark it down for being expensive on intrinsic worth, and two further things argue for patience: the model's own twelve-week projection sits slightly below today's price, and the company reports half-year results tomorrow. We wait for the results and a stronger signal. OPS: cv70 PullbackBuy_VWAP (L1-favoured), BuyRank #3, eD -0.50, Quality_Pass TRUE so the gate opens. val_adj -10 (MoS -2.054) -> adj 60 vs the 65 floor. Est_12W 3.92 is BELOW the 3.96 close (-1.0pct) - a negative forward view on a NEW entry. F34 1H 2026 results 08-12. size_mod would have been 0.75x (-3 < MoS <= -1). RE-ARM on a cv>=80 reprint post-results with a non-negative Est_12W. Caveat logged: PullbackBuy_VWAP|buy is 0/4 on 1wk targets in this book - sample too small to bias, printed not applied. |
| S68 | Singapore Exchange | 25.20 | HOLD | 25.20 | 24.01 | 25.20 (+0.0%) | 24.97 (-0.9%) | 24.98 (-0.9%) | The exchange closed at the highest price of the period on healthy volume and the model refused to evaluate it for a third straight session, citing a liquidity check that is plainly misfiring on an active, liquid stock. We cannot act on a signal that will not compute, so we do the one thing that is unambiguously right: raise the trailing stop under a position that is making highs. OPS: 3rd consecutive NO TRADE cv0 on 'Liquidity gate failed', this time on a 25.20 window-high close with 4.44m sh traded. Stop RAISED 23.56 -> 24.01 (2.0 x ATR 0.5938 chandelier off 25.20). See systemic finding 3 - the gate has now nulled three legitimate evaluations in a row including O39 tonight on 13.25m shares. Singtel Q1 update 08-18 reads across via the index. |
| 9CI | CapitaLand Investment | 2.72 | HOLD | 2.72 | 2.62 | 2.71 (-0.4%) | 2.73 (+0.4%) | 2.82 (+3.7%) | Upgraded to a hold with the healthiest twelve-week outlook among our property holdings, and it reports half-year results in two sessions. Our policy through a scheduled announcement on a position we already own is to hold and trail, never to pre-trim into it. No action. OPS: HOLD cv60 FailedBreakdown_Up_20D, Est_12W 2.82 = +3.7pct - best forward view in the held REIT/property complex. adj 50 (val_adj -10, MoS -1.916) is well under the floor for an add. Stop HELD at 2.62; the live model stop 2.60 is lower and we never lower a stop. 1H results 08-13 -> catalyst-aware hold. |
| A17U | CapitaLand Ascendas REIT | 2.51 | HOLD | 2.51 | 2.49 | 2.51 (+0.0%) | 2.53 (+0.8%) | 2.53 (+0.8%) | No trade signal and no action needed on the position itself, but a bookkeeping matter that matters: the trust goes ex-distribution on 13 August, which mechanically marks the quoted price down about three percent. We move the risk line down by the same amount on that date so the drop is not mistaken for a genuine break, which is exactly the error that produced a phantom alert on another holding earlier this month. OPS: MANDATORY REBASE - 7.482c ex-distribution on 08-13 = -2.98pct of a 2.51 close. Stop MUST move 2.49 -> 2.42 that morning; left alone the row prints ~2.44 and trips a false breach (identical mechanism to B61 on 08-05). NO TRADE cv0 'Consensus conflict + low conviction'; adj n/a. |
Outcome & track record
Accuracy at the time of this record.
109 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 →
0.229 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.