Singapore — End-of-Day Verdict
32nd cycle: the buy pool goes to **absolute zero** for the first time in the cycle as G07 reverts BUY cv71 → NO TRADE cv0 in one session and a bank-led risk-off tape takes names down; **U11's unexecuted 08-17 trim is re-published unchanged as SG-1 cycle 2** with both L2 limbs firing *deeper* than last night; J69U prints its **loudest sell ever (STRONG SELL cv81, Sell_Rank #1)** and SG-2 refuses it for the 12th time on both limbs; and the 08-17 EOD is discovered to have written ledger + state but **no memory block at all**
Recommendations
0 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 | 40.88 | SELL 30% | 40.88 | 41.27⚠ | 40.80 (-0.2%) | 40.81 (-0.2%) | 40.60 (-0.7%) | UOB has now closed below its dividend-adjusted stop for a fifth straight session while the model's own 12-week view has turned negative for a second day, so we are stepping 30% out of the book's second-largest holding rather than waiting for a sell label the engine is not going to print. The bank is the most rate-exposed of the three Singapore lenders (roughly 70% of revenue from net interest income versus ~60% at DBS and OCBC), it has already trimmed its interim dividend, and analysts have stayed guarded since the 1H2026 result on guidance that looks weaker than peers; with FOMC minutes tomorrow and the new Fed chair's first Jackson Hole keynote on 28 Aug, cutting the largest rate-sensitive exposure ahead of those prints is cheap insurance. Valuation is mildly rich (OVERVALUED_VS_RI, MoS -0.55) but is amplify-only and not extreme enough to bump the tier. | L2/CLAUDE.md-9 ORIGINATION, not a signal label: close 40.88 vs the 88c-rebased stop 41.27 (-0.94%, 0.62 ATR, 5th consecutive session below) AND Target_12Week 40.60 vs close 40.88 (-0.68%, negative for a 2nd session after -0.17% on 08-17). The stop-break + negative-12wk combination is itself the signal-based reason and bypasses the 65 conviction floor, which governs signal-label candidates only. Engine prints NO TRADE cv0 'Consensus conflict + low conviction' for a 4th consecutive settle and all 5 intraday prints today. SG-1 CYCLE 2 - this is a re-publication of the 08-17 card (same, same market order) that was never executed because the holdings file is byte-identical since 2026-07-15; SG-1 escalates a stalled de-risk to an unconditional market exit at the next open, which this already is. SIZE NOT RE-ESCALATED (SG-8): 30% minimum tier held even though the NoTrade|sell haircut cleared - this run (n=8, hit_1w 0.50 but hit_4w still 0.00) - a stalled de-risk is re-published at its original size, never bigger. If this is not executed by the 08-19 close it becomes SG-8 cycle 3 and converts to a blocking holdings-sync flag. ACTION-PRICE PROVENANCE: the NO TRADE row prints no Entry_Price, so this is a MARKET order referenced to the anchor settled close 40.88 - nothing carried forward, nothing invented. Estimates vs 40.88: 1wk 40.80 [39.89-41.92] -0.20%, 4wk 40.81 [38.03-43.57] -0.17%, 12wk 40.60 [36.14-45.49] -0.68% - a flat-to-down path, which is exactly what the trim steps out of. Proceeds ~S; residual (S) keeps the 41.27 stop and drops concentration 26.9% -> 18.7% of the book. · news |
| J69U | FRASERS CPT TR | 2.18 | HOLD | 2.19 | 2.18 | 2.19 (+0.0%) | 2.19 (+0.0%) | 2.22 (+1.4%) | Frasers Centrepoint printed its loudest sell signal ever tonight, and we are declining it for the twelfth time. The trust closed exactly on its stop rather than through it, the model still projects the price higher over 12 weeks, and the business itself is improving - earnings forecast to compound ~8.4% a year with net margins up from 49.3% to 51.2%. Singapore sell labels on this book reverse more than half the time, and this exact name flipped from STRONG SELL back to HOLD only one session ago. | SG-2 CONFIRMATION FAILS ON BOTH LIMBS: close 2.18 vs recorded stop 2.18 is touching, NOT a decisive break; Est_12W 2.22 is +1.83% POSITIVE. Signal family is new (BreakdownDown_20D, n=2 report-only, no bias) and conviction 81 clears the floor on arithmetic, but SG-2 is the governing veto. Intraday path SSELL23 -> SELL58 -> SELL58 -> SSELL81 -> SSELL81. Printed model stop 2.12 is a bearish-family short-side level and is REJECTED - never lower a recorded stop. This would have been the board's largest sell at 80.07% x 28,000 = = SRE-ARM: a decisive close below 2.18 on a NEGATIVE Est_12W. · news |
| A17U | CapitaLand Ascendas REIT | 2.43 | HOLD | 2.43 | 2.42 | 2.43 (+0.0%) | 2.46 (+1.2%) | 2.46 (+1.2%) | Ascendas REIT has printed a strong sell label for three sessions running, but it is still trading a cent above its stop, the model projects it higher over both 4 and 12 weeks, and the prevailing analyst read is that the units trade at a discount to fair value after the April raise brought leverage down. Holding. | SG-2 FAILS ON BOTH LIMBS: close 2.43 vs the DPU-rebased 2.42 stop is NO break; Est_12W 2.46 is +1.23% POSITIVE. Arithmetic clears the floor (70 raw - 3 RallySell_BearTrend|sell ACTIONABLE bias = 67) but SG-2 vetoes. MoS -7.685 is deeply negative and would amplify a sell, but valuation NEVER originates one (CLAUDE.md-9) and there is no signal-based case here to amplify - SG-3 says read the magnitude, not the REIT band, and the magnitude has no trigger to attach to. Trade would have been trivial anyway: 80.07% x 400 = = SRE-ARM: decisive close below 2.42 on a NEGATIVE Est_12W. · news |
| O39 | OCBC Bank | 30.95 | HOLD | — | 30.09 | 30.78 | 30.50 | 31.29 | The OCBC profit-take is still valid and still unexecutable. The position is up ~19% on a real, the signal collapsed outright today, and the 12-week view is dead - but this card has been published five times against a holdings file that has not changed since 15 July, so it is not re-published again. | SG-8 CYCLE 5. ADR-0012 fire remains valid on the book's ONE genuinely armed real-cost position (Avg_Cost 25.9325, +19.35%, Gain_ATR 7.16, T12_Progress 98.91): tick (b) FIRES - conviction collapsed, a -78 pt drop against a -20 threshold; tick (c) FIRES - dead 12wk, Target_12Week 31.29 < 1.02 x 30.95 = 31.57. Ticks (a) and (d) do not fire (PROFIT_ARMED is the only warning code; only 1 NO TRADE print). Mechanically 2 ticks = profit_take_2 = 50%, up from the 30% published 08-13/08-14. Per SG-8 the card is NOT re-published and the size is NOT re-escalated. Counterfactual: the unexecuted trim has now started to cost - O39 was 31.50 when first re-published 08-13 and closed 30.95 tonight (-1.75%), so the delay is no longer free. Unblocks the moment input/sgportfolio.json is synced. |
| B61 | BT SEMBAWANG EST | 4.46 | HOLD | — | 4.62⚠ | 4.46 | 4.46 | 4.54 | Bukit Sembawang has closed below its stop for eight straight sessions on a S position, but the model still projects it higher over 12 weeks and the valuation layer rates it undervalued, so the exit case has never actually completed. | SG-8 CYCLE 4. L2 does NOT fire: Est_12W 4.54 is +1.79% ABOVE close 4.46 - the 12wk limb has failed every session since the card was first raised (contrast U11, where it crossed). RSI 31.3 is oversold. Blocked pending holdings sync; no re-publication, no size change. |
| HMN | CAPITALAND and ASCOTT STAPLE U | 0.86 | HOLD | 0.86 | 0.88⚠ | 0.86 (+0.0%) | 0.86 (+0.0%) | 0.88 (+2.3%) | Ascott Trust is in its tenth session below stop but the label has faded from STRONG SELL back to AVOID, the 12-week view is positive, and the position is worth S - it is noise on this book. | SG-8 CYCLE 8. Label decayed STRONG SELL cv71 (08-17) -> AVOID cv60. Est_12W 0.88 is +2.92% positive, so SG-2 fails independently of the block. FailedBreakout_Down_20D|sell bias is 0 (n=73, hit.507) - a coin flip. |
| BUOU | FRASER L&C TRUST | 0.95 | HOLD | — | 0.96⚠ | 0.95 | 0.95 | 0.97 | Frasers Logistics broke its stop decisively for the first time tonight on the book's fourth-largest position, but the model still sees it higher in 12 weeks, so the exit combination is one limb short. It is the name most likely to become the next disposal. | FIRST DECISIVE BREAK: close 0.950 vs stop 0.960 (-1.04%) after touching exactly 0.960 on 08-17. L2 does NOT fire - Est_12W 0.97 is +2.11% POSITIVE. NO TRADE cv0, RSI 31.3, Sat risk. MoS -6.033 is amplify-only with no signal to amplify (SG-3). TRIGGER: a second decisive close below 0.96 on a NEGATIVE Est_12W makes BUOU the immediate lead dispose. |
| ME8U | Mapletree Industrial Trust | 1.91 | HOLD | — | 1.89 | 1.91 | 1.90 | 1.89 | Mapletree Industrial is the mirror image of Frasers Logistics - its 12-week view has gone negative but the price is still above its stop, so it is the other name one limb away from a disposal. | Est_12W 1.89 vs close 1.91 = -1.05% NEGATIVE (the only held name besides U11 with a negative forward view), but close 1.91 > stop 1.89 so there is no break. Flipped to STRONG SELL - REDUCE cv34 intraday at 16:00/16:30 and reverted to NO TRADE cv0 at the settle - L4 makes the settle authoritative. Sthe book's third-largest position. TRIGGER: a decisive close below 1.89 completes the L2 combination. |
| N2IU | Mapletree PanAsia Com Tr | 1.26 | HOLD | 1.26 | 1.27⚠ | 1.26 (+0.0%) | 1.27 (+0.8%) | 1.28 (+1.6%) | Mapletree PanAsia slipped a cent below its stop but the 12-week view is positive and the position is S - hold. | Close 1.26 vs stop 1.27 (-0.79%) after reclaiming it on 08-14. Est_12W 1.28 is +1.59% positive, so L2 does not fire. AVOID cv50 at the settle after two STRONG SELL intraday prints (L4). Printed model stop 1.31 is a bearish-family short-side level - REJECTED, never raise a stop above the close on a long. |
| D05 | DBS Group Holdings | 75.72 | HOLD | 75.72 | 74.47 | 75.44 (-0.4%) | 75.17 (-0.7%) | 77.61 (+2.5%) | DBS gave back 1.5% in the bank-wide de-rate but still carries the best forward view of the three lenders and sits above its trailing stop. Hold, no add. | Close 75.72, only 1.68% above the 74.47 trail after a -1.51% session. Est_12W 77.61 = +2.50%, the best among the banks. Conviction fading -> across four settles. SG-4 governs bank cores - no chase, add only on a non-extended BUY-ADD reprint at cv>=65. Chandelier computes 73.78, BELOW the recorded 74.47 - trail HELD, never lowered. |
| S68 | Singapore Exchange | 25.01 | HOLD | — | 24.44 | 25.11 | 24.99 | 24.87 | SGX fell 1.9% from its fresh high and the signal has gone flat - the decision to pull the 24.27 add-queue on 08-17 rather than chase the print looks right two sessions later. Hold on the trail. | NO TRADE cv0 (from HOLD 58). Close 25.01 vs trail 24.44, 2.3% of room. Chandelier from the 25.30 high computes 24.20, BELOW the recorded 24.44 - trail HELD. The withdrawn 24.27 queue never filled and the low since has been 24.82, so the no_chase_queue counterfactual stays open on this name. Est_12W 24.87 is -0.56% vs close - the model has not endorsed the recent range for weeks. |
| S63 | ST Engineering | 11.15 | HOLD | 11.16 | 10.55 | 11.12 (-0.4%) | 11.17 (+0.1%) | 10.92 (-2.2%) | ST Engineering was one of only five names green today and prints a bullish anticipation setup, but it is a held name whose own 12-week estimate sits below the current price, so the answer is a tighter stop rather than more stock. | HOLD cv70, AnticipationUp_VCP with a Buy Stop @11.16 - but Action is HOLD, not BUY-ADD, so it is NOT in the buy pool. Est_12W 10.92 is -2.06% BELOW close 11.15. Whipsawed hard intraday (AVOID50 -> HOLD68 -> SSELL58 -> SSELL58 -> HOLD70); L4 takes the settle. Trail RAISED 10.32 -> 10.55 (live model stop). 4c ex-dividend 08-25 - rebase the stop to 10.51 on the day. |
| U13 | UOI LTD | 8.60 | HOLD | — | 8.03 | 8.61 | 8.64 | 8.64 | UOI fell 1.7% with the financial complex but remains well above its stop with a flat forward view. No signal, no action. | NO TRADE cv0, close 8.60 vs stop 8.03. Est_12W 8.64 = +0.47%. Illiquid - the scan has repeatedly gated it out on the liquidity screen. |
| C07 | Jardine C&C | 27.42 | HOLD | 27.42 | 26.38 | 27.59 (+0.6%) | 28.01 (+2.2%) | 28.16 (+2.7%) | Jardine Cycle & Carriage remains the only name in the book that is cheap, passes the quality screen and is flagged a bargain - and it still has no signal to act on. Watch. | UNDERVALUED, MoS +0.628 (val_adj +10, the maximum), Quality_Pass TRUE, EPV 73.39 vs close 27.42. But the label is AVOID cv43 -> adj 53, below the 65 floor. Est_12W 28.16 = +2.70%. Position is only (S). RE-ARM on any BUY-family reprint at cv>=55. |
| 9CI | CapitaLand Investment | 2.67 | HOLD | — | 2.62 | 2.67 | 2.69 | 2.79 | CapitaLand Investment was the worst held performer today (-2.55%) and now sits five cents above its stop, but the 12-week view is the strongest in the book. Hold. | NO TRADE cv0, close 2.67 vs stop 2.62. Est_12W 2.79 = +4.49%, the best forward view of any held name. Watch for a non-extended BUY-ADD reprint. |
| S58 | SATS | 4.76 | WATCH | 4.76 | 4.67 | 4.76 (+0.0%) | 4.76 (+0.0%) | 5.10 (+7.1%) | SATS has the best forward shape of any unheld name and firmed on a down day, but it is still only a WATCH and the valuation haircut pushes it under the bar. No entry. | WATCH cv63 (rising ->, FailedBreakdown_Up_20D - the family L1 explicitly discounts as a chase. Not in the buy pool (WATCH is not an actionable bullish label). Fusion would be 63 - 10 (MoS -2.346 x 5, clamped) = 53, below the 65 floor. Est_12W 5.10 = +7.14%. RE-ARM on a BUY-family reprint at cv>=75 given the -10 valuation haircut. |
| G07 | Great Eastern | 22.55 | WATCH | — | — | 22.50 | 22.13 | 22.62 | Last night's quality-gate rejection of Great Eastern is vindicated one session later - the BUY signal has already evaporated and the stock is down. No entry. | BUY cv71 Buy_Rank #1 (08-17) -> NO TRADE cv0 (08-18), price -0.84% to 22.55. The 08-17 reject rested on Quality_Pass FALSE (F_Score 3/9, Quality_Rank 16/26, Earn_Yield -2.26%), RSI 81.9, and all three estimates below the buy-stop; RSI is still 77.3 and Est_4W 22.13 is -1.86% below close. 35c ex-dividend 08-19 is a further mechanical -1.5%. L7: the quality gate blocks NEW entries only, so this does not govern sgotr, which already holds G07. |
Outcome & track record
Accuracy at the time of this record.
117 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.