Singapore — End-of-Day Verdict
38th cycle: **the sell side finally clears the bar too** — after five consecutive zero-dispose EODs, **U11 completes the L2 combo on a S position** (8th straight close below the 41.27 trail — and the trail rebase is provably the Sdividend, so the breach is *clean price weakness*, not the ex-div arithmetic that has excused B61 and HMN for weeks) with **all three forward horizons simultaneously negative**, taking a 30% trim; **S68 ramps again to `cv100`** — the model's ceiling — and reprices its unfilled queue to 24.80 at adj 91; and **O39's profit-take finally fires** on a −53pt conviction collapse, only for SG-8 to correctly refuse to re-publish it, because the identical card from 08-21 has been sitting unexecuted for a 29th static cycle
Recommendations
1 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 |
|---|---|---|---|---|---|---|---|---|---|
| U11 | United Overseas Bank | 40.95 | SELL 30% | 40.95 | 41.27⚠ | 40.81 (-0.3%) | 40.90 (-0.1%) | 40.72 (-0.6%) | UOB has now closed below its risk line for eight straight sessions and its whole forward curve has turned negative, so we are taking 30% off the book's largest position while the fundamentals still look good on paper. Q2 net profit rose 10% to S48bn with record wealth-management fees, and the bank is selling UOB Asset Management to Allianz GI for S - yet the shares have drifted lower through every one of those headlines. A stock that will not rise on good news is telling you something. On our own valuation work UOB screens as overvalued against its residual-income anchor and it carries the weakest quality score in the Singapore book (F-Score 4, quality rank), so this is the position we would rather own less of going into a two-way macro week. We are trimming, not exiting - the remaining 70% keeps full exposure to the dividend and any re-rating. OPERATIONAL: L2 origination (decisive-by-persistence close below the recorded trail 41.27 since 2026-08-17, 8 consecutive sessions, post-ex-dividend rebase 42.15->41.27 = exactly the Sdividend so the breach is clean price weakness, NOT distribution arithmetic) AND all three Est_* horizons simultaneously negative (1W -0.34%, 4W -0.12%, 12W -0.56%) - the same configuration on which the 08-17/08-18 SELL 30% card was originally raised before being withdrawn 08-21 on a limb flip; the flip has reversed with 8 sessions of persistence behind it now. Conviction floor 65 does not apply - L2 is an explicit signal-based exit basis independent of the label (engine prints NO TRADE cv0, 9th consecutive session, blank Stop_Price). Trim-discipline gates G1/G2/G3 and the catalyst-aware-hold clause are OUT OF SCOPE: they are scoped to profit-trims on uptrending names; U11 is neither (PROFIT_ARMED is spurious per SG-7 - no Avg_Cost; price is below EMA20, RSI 46.9). The block's own Contrast line exempts a decisive stop-loss break. Valuation amplify: MoS_FV -0.507 is not <= -3, so no tier bump - minimum 30% tier stands. Scorecard NoTrade|sell actionable +1 (n=9) but note hit_4w = 0.000 on that cell - the 1wk read is the reliable leg, the 4wk is not. SG-8 does NOT suppress this: the prior U11 SELL 3,700 card was formally WITHDRAWN 2026-08-21, so this is a fresh card, not a re-publication. Event Horizon (advisory): executes at the 08-27 open, ahead of Jackson Hole 08-28 (Warsh's first keynote as Fed chair, bank leg) and the MSCI August review in the 08-31 closing auction - de-risking INTO a two-way event on a name already failing its trail is the right direction. Residual keeps stop 41.27 tonight (unchanged until the fill); re-base the residual trail at the next EOD post-execution. Position is 27.2% of - NOT a sizing origination (CLAUDE.md S9 forbids it), stated for context only. · news |
| S68 | Singapore Exchange | 25.57 | ADD — QUEUED @24.80 | 24.80 | 24.06 | 25.45 (+2.6%) | 25.24 (+1.8%) | 25.71 (+3.7%) | SGX is the strongest thing in the Singapore book and it just printed the highest conviction score the model produces. The exchange is compounding on a genuine structural tailwind - capital is rotating into Asia, securities turnover and derivatives volumes are surging, and the IPO pipeline has reopened after years of drought - which is why the shares are up roughly 36% this year with a 39% one-year total return. We want more of it, but not at any price: the stock closed 2.1 ATR above the model's buy level, so we are placing a good-till-cancelled limit 2.9% below the close rather than chasing. If it pulls back to us we add; if it runs away we keep what we own. Note the shares screen expensive against every intrinsic anchor we compute, which is why the order is deliberately small. OPERATIONAL: conviction ramp -> across four snapshots, monotonic, Buy_Rank #1, on PullbackBuy_VWAP - the family L1 favours. Action price 24.80 = the anchor snapshot's live Entry_Price exactly (provenance rule); the 08-25 queue @24.72 never filled (session low 25.30) and is REPRICED, not carried. Entry range 24.71-24.89 (+-0.25 ATR, ATR 0.370). Extended: Entry_Distance_ATR -2.08 -> queue, never a market fill. SG-6 cv>=95 immediate-tranche carve-out is TRIGGERED on conviction (cv100) but BLOCKED by its own extended/RSI-hot exclusion (eD -2.08 plus a fresh RSI_DIV print) - strict no-chase stands, pure limit, no tranche. RSI_DIV on the row does NOT make S68 a trim candidate: the valuation-elevate path needs MoS_FV < -3 and S68 is -1.81; PROFIT_ARMED is spurious per SG-7 (no Avg_Cost), so no profit-take path exists either. Sizing chain: Portion 114.38% x 3,000 = -> x0.75 size modulator = 2,573 -> max_position_pct headroom (60% x S base - Sexisting) = 1,745 -> CAP S / 24.80 = 970 -> (SGX 100-lot). Cost Sleaving S., not the model, is the binding constraint. Stop: queued tranche carries the model's live Stop_Price 24.06; the HELD position's trail STAYS 24.44 - a trail is never lowered - and the blended stop re-derives at the next EOD on a fill. Scorecard: PullbackBuy_VWAP|buy is n=4, hit_1w 0.000 - BELOW the n>=5 floor, report-only, bias 0 applied; the book's favourite buy family is now 0-for-4 on 1-week targets and only the sample floor keeps that out of the arithmetic. This remains the single largest risk on the card. 12wk estimate 25.71 far exceeds the DCF anchor 17.97 and the ResInc FV 8.86 - projection assumes growth well beyond what is priced in; treat the upper band as aspirational. Event Horizon (advisory): NVIDIA Q2 FY2027 prices into SGX on 08-27 and the MSCI August review lands in the 08-31 closing auction - a risk-off gap could fill this limit, which is exactly what the limit is for. · news |
| O39 | OCBC Bank | 31.44 | BLOCKED - sync holdings | — | 30.42 | 31.29 | 31.33 | 32.12 | OCBC's profit-take trigger has finally fired, but the trim we already asked for on 21 August has never been executed, so there is nothing new to decide - the existing instruction to sell simply stands, and tonight's signal makes it more urgent rather than less. The bank itself is doing well (record first-half profit up 13%, interim dividend lifted 15%), but the position is up 21% on cost and the model's read on it collapsed today. Nothing further will be published on OCBC until the holdings file reflects the sale. OPERATIONAL: ADR-0012 profit-take FIRES for the first time since 08-21. Armed per SG-7 on a REAL Avg_Cost 25.93 (+21.24%, Gain_ATR 10.31, T12_Progress 97.88). Tick (b) conviction collapse: = -53 pts, far beyond the >=20 threshold - FIRED. Tick (c) dead-12wk: Target_12Week 32.12 vs 1.02 x 31.44 = 32.0688 - clears by S, a SECOND consecutive four-cent miss (was Son 08-25); no hysteresis on this tick, see systemic finding. Tick (a) no other Exit_Warning code (PROFIT_ARMED alone). Tick (d) only 1 NO TRADE print (T3 was HOLD) - needs 2 consecutive, arms tomorrow. 1 tick -> profit_take_1 = 30% =, which is EXACTLY the outstanding 2026-08-21 card. G1 softening irrelevant - tick (b) fires regardless. Valuation amplify: MoS_FV -0.745 not <= -3, no tier bump. SG-8 GOVERNS: the 08-21 SELL card has been carried 4 cycles with the monitor reporting no change - do NOT re-publish the card, do NOT re-escalate size, emit the single blocking sync flag. Tonight's fire RE-VALIDATES the outstanding card rather than creating a new one. Trail held at 30.42 (model prints no stop under NO TRADE). Clear with 'trade sell sg O39 300' or 'trade skip sg O39'. · news |
| B61 | BT SEMBAWANG EST | 4.33 | HOLD - SG-2 refusal | 4.33 | 4.62⚠ | 4.33 (+0.0%) | 4.34 (+0.2%) | 4.41 (+1.8%) | Bukit Sembawang printed the loudest sell label on the board tonight, and we are declining it for the same reason we declined it last week: the forward view is still positive and much of the apparent damage is a dividend that came out of the price, not value that was lost. The stock is deeply oversold, it trades below our estimate of what the land bank is worth, and the model's own view twelve weeks out sits about 2% above tonight's close. We hold. OPERATIONAL: STRONG SELL - REDUCE cv37, Sell_Rank #1, Portion 80.07%, BreakdownDown_20D. SG-2 leg 1 (stop break) PASSES - 15th consecutive breach, close 4.33 vs recorded 4.62 = -6.28%; but note the close sits ABOVE the model's own live Stop_Price 4.23. SG-2 leg 2 (negative 12wk) FAILS - Est_12W 4.41 vs 4.33 close = +1.85%. Refused. Corroborating: label whipsawed AVOID 47 -> SELL-REDUCE 70 -> HOLD 57 -> STRONG SELL 37 across four snapshots - unstable, and the cv37 on a STRONG SELL label is itself a contradiction; RSI 21.1 deeply oversold with ReversalUp_Oversold printed only yesterday. Ex-dividend confirmed: SGD 0.22 ex-date 2026-08-05 on a ~4.84 stop = the 4.62 rebase, corroborating the 08-24 finding that a large share of the breach is distribution arithmetic. Valuation ARGUES DOWN: UNDERVALUED, MoS_FV +0.261 - selling cheap, trim the minimum if ever. Scorecard BreakdownDown_20D|sell n=2, report-only, no bias applied. · news |
| Z74 | Singtel | 4.51 | NO ACTION - declined | 4.51 | 4.48 | 4.51 (+0.0%) | 4.54 (+0.7%) | 4.43 (-1.8%) | Singtel was the second-strongest buy signal tonight and we are passing on it. The technical setup is clean - the entry sits right at tonight's close, so there would be no chasing involved - but the model's own twelve-week view is BELOW where the stock trades today, and we do not open a new position whose best forward estimate is negative. The fundamentals back that up: first-quarter net profit fell 72% to S, the Singapore mobile business shrank 4.6% on competition, and the 9% dividend increase is being funded by selling assets (the S5bn Airtel stake) rather than by earnings. Cheap on a headline multiple, but the earnings are going the wrong way. OPERATIONAL: BUY cv81, Buy_Rank #2, FailedBreakdown_Up_20D, Entry_Distance_ATR 0.0 (genuinely non-extended, unlike S68), RSI 61.1, conviction ramp ->. Adjusted conviction 81 - 3 + 0 = 78, CLEARS the 65 floor. Quality gate PASSES (Quality_Pass True, F-Score 5) so the hard gate does not block this new entry. Declined on three grounds: (1) Est_12W 4.43 vs 4.51 close = -1.8% - the same forward-view ground on which BN4 was declined 08-24 and 08-25, applied consistently; (2) L1 explicitly discounts FailedBreakdown_Up_20D chases (~0.25 hit, stop-first ~60%) and the cell is UNGRADED on this book - no scorecard entry, report-only, so the family gets no empirical support here; the model stop 4.48 is only 0.42 ATR below entry, i.e. very easily taken out. (3) : is committed to the higher-ranked S68 card (adj 91 vs 78), leaving S - below a sensible tranche. Grounds (1) and (2) stand independently of (3). L7 NOTE: Z74 is also carried in the sgotr sub-book - sg is the governing parent, execute once. · news |
| D05 | DBS Group Holdings | 76.24 | HOLD - trail 73.04 | 75.17 | 73.04 | 76.05 (+1.2%) | 76.30 (+1.5%) | 78.83 (+4.9%) | DBS holds. No stop break (close 76.24 well above trail), positive 12-week view (+3.4%), and the signal is the constructive PullbackBuy family. OPERATIONAL: HOLD cv53, no sell-pool entry. Trail RAISED 74.47 -> 75.17? NO - model live Stop_Price is 73.04, BELOW the existing 74.47 trail; a trail is never lowered, so the recorded stop STAYS 74.47. Event Horizon: Jackson Hole 08-28 bank leg + MSCI review 08-31, both advisory. |
| ME8U | Mapletree Industrial Trust | 1.93 | HOLD - no stop break | 1.93 | 1.89 | 1.93 (+0.0%) | 1.92 (-0.5%) | 1.90 (-1.6%) | Mapletree Industrial holds - no stop break, stable HOLD for a third session. OPERATIONAL: HOLD cv53, stop 1.89 vs close 1.93. Est_12W 1.90 vs 1.93 = -1.6% NEGATIVE, but leg 1 of L2 (stop break) FAILS, so no exit basis - both legs required. Watch: if price breaks 1.89 the L2 combo completes. Event Horizon: NVIDIA Q2 prices into ME8U's data-centre portfolio on 08-27 (advisory). |
| BUOU | FRASER L&C TRUST | 0.94 | HOLD 170,198 - 7th break | — | 0.96⚠ | 0.94 | 0.94 | 0.96 | Frasers L&C holds despite a 7th consecutive session below its stop, because the twelve-week view remains positive. OPERATIONAL: 7th stop breach (close 0.94 vs 0.96 = -2.08%). L2 leg 2 FAILS - Est_12W 0.96 vs 0.94 = +2.1% positive. NO TRADE cv0 for a 4th session, NoTrade_Reason 'Consensus conflict + low conviction'. Contrast with U11 tonight (negative curve -> acted): the two-leg test is doing real work in both directions. |
| J69U | FRASERS CPT TR | 2.18 | HOLD 28,000 | 2.18 | 2.23⚠ | 2.18 (+0.0%) | 2.19 (+0.5%) | 2.21 (+1.4%) | Frasers Centrepoint's sell signal has evaporated within one session, vindicating last night's refusal to sell. We hold. OPERATIONAL: STRONG SELL - REDUCE cv71 Sell_Rank #1 (08-25) -> AVOID cv43 (08-26). Falls OUT of the sell pool entirely one session after SG-2 refused it for the 14th time - the refusal is vindicated inside a single session on a Sposition. Close 2.18 sits BELOW the recorded trail 2.18? Equal, not below - no decisive break; Est_12W 2.21 vs 2.18 = +1.4% positive. SG-2 discipline continues to govern; this is the clearest single-session validation the rule has produced. |
| C07 | Jardine C&C | 27.20 | HOLD - trail 26.56 | 27.20 | 28.21⚠ | 27.28 (+0.3%) | 27.53 (+1.2%) | 27.40 (+0.7%) | Jardine C&C holds - AVOID is not a sell label, the forward view is positive, and it is the only genuinely cheap, quality-passing name in the book. OPERATIONAL: AVOID cv51 on RallySell_BearTrend - NOT in the sell pool. Scorecard RallySell_BearTrend|sell is actionable NEGATIVE at -3 (n=9, hit_1w 0.333) - the cell argues against acting on this family at all. Est_12W 27.40 vs 27.20 = +0.7%. UNDERVALUED, MoS_FV +0.621, Quality_Pass True, Quality_Rank #3. Model stop 28.21 is ABOVE the close (bearish-family artifact) - recorded trail stays 26.56, never raised above the price. |
| N2IU | Mapletree PanAsia Com Tr | 1.27 | HOLD 10,000 | 1.27 | 1.32⚠ | 1.27 (+0.0%) | 1.28 (+0.8%) | 1.27 (+0.0%) | Mapletree PanAsia sits exactly on its stop with a flat forward view - not a break, not a sell. Hold. OPERATIONAL: close 1.27 = recorded stop 1.27, breach 0.00% - NOT a decisive break, L2 leg 1 fails. Est_12W 1.27 = close, flat, not negative - leg 2 also fails. AVOID cv43, not in the sell pool. First time this name has touched its stop; a decisive close below 1.27 with the curve negative completes the L2 combo. |
| S63 | ST Engineering | 10.58 | HOLD - trail 10.51 | — | 10.51 | 10.62 | 10.58 | 10.44 | ST Engineering holds above its stop. OPERATIONAL: NO TRADE cv0 (reverted from HOLD cv53 on 08-25), close 10.58 vs trail 10.51 - above, no breach. Est_12W 10.44 vs 10.58 = -1.3% negative, but leg 1 (stop break) fails so no L2 basis. Watch the 10.51 line - a close below it completes the combo. |
| U13 | UOI LTD | 8.70 | HOLD - liquidity gate | — | 8.03 | 8.69 | 8.74 | 8.71 | UOI holds - the model cannot trade it. OPERATIONAL: NO TRADE cv0, NoTrade_Reason 'Liquidity gate failed' verbatim, 4th consecutive session; volume on the session. Close 8.70 well above trail 8.03. Forward curve flat-positive. Untradeable size on this book regardless of signal. |
| A17U | CapitaLand Ascendas REIT | 2.45 | HOLD 400 | 2.45 | 2.51⚠ | 2.45 (+0.0%) | 2.45 (+0.0%) | 2.44 (-0.4%) | CapitaLand Ascendas holds - AVOID is not a sell, and the position is a rounding error at SOPERATIONAL: AVOID cv60, out of the sell pool for a 4th session (was STRONG SELL cv71 on 08-21). Est_12W 2.44 vs 2.45 = -0.4%, marginally negative, but no stop break (2.45 vs trail 2.42) so L2 leg 1 fails. MoS_FV -7.757 is extreme BUT per SG-3 the SG REIT valuation band is miscalibrated and valuation never originates a sell - it could only amplify a signal-based case, and there is none. Position = Simmaterial. |
| HMN | CAPITALAND and ASCOTT STAPLE U | 0.86 | HOLD - ex-div rebase flagged | 0.86 | 0.88⚠ | 0.86 (+0.0%) | 0.86 (+0.0%) | 0.89 (+3.5%) | CapitaLand Ascott holds - the 16th straight 'breach' is almost certainly an unrebased distribution, and the forward view is positive. OPERATIONAL: 16th consecutive stop breach (0.86 vs 0.88 = -2.27%), flagged on 08-24 as PROBABLY ENTIRELY unrebased distribution arithmetic. L2 leg 2 FAILS - Est_12W 0.89 vs 0.86 = +3.5% positive. AVOID cv53, not in the sell pool. Position Simmaterial. Operational follow-up: the 0.88 stop needs a distribution rebase or this name flags the monitor indefinitely. |
| 9CI | CapitaLand Investment | 2.68 | HOLD 2,000 - best 12wk | 2.68 | 2.57 | 2.68 (+0.0%) | 2.69 (+0.4%) | 2.74 (+2.2%) | CapitaLand Investment upgrades from no-signal to HOLD and carries the best forward view in the book. OPERATIONAL: NO TRADE cv0 x3 -> HOLD cv54 on FailedBreakdown_Up_20D. Est_12W 2.74 vs 2.68 = +2.2%, the strongest 12-week limb of any held name. Trail RAISED 2.62 -> 2.57? NO - model stop 2.57 is BELOW the existing 2.62 trail; a trail is never lowered, recorded stop STAYS 2.62. No stop break. |
Outcome & track record
Accuracy at the time of this record.
119 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.