This website is for educational purposes only and does not constitute investment advice. Always do your own research and assess your own risk tolerance before making investment decisions.

SG End-of-Day Verdict 26 Aug 2026, 17:00:00 GMT+8

SG Mid/Small Cap — End-of-Day Verdict

NO ACTION

Recommendations

0 to acquire · 0 to dispose

TickerNameCloseActionAction 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
PCTPC Partner Group2.78WATCH2.952.783.07 (+4.1%)3.35 (+13.6%)3.24 (+9.8%)PC Partner gave us the setup we have been waiting five weeks for and it landed three points short. It printed the highest-quality buy signal our system recognises, on the top-ranked row in the book, on a day it finally traded calmly enough for us to see it at all - and the fresh valuation work published tonight flipped it back to cheap, at a 7.7% discount to what we think the business is worth, with the second-best quality score of the eleven names. Our twelve-week view is the best in the book at +16.5%. What is missing is conviction: the signal scored 60 where we require 63 on a name with this valuation help. We do not bridge a three-point gap by hand, and the entry the model wants is 6.1% ABOVE today's price - a breakout-stop, not the cheap pullback the same signal offered on 6 August - so patience costs us little. OPERATIONAL: BUY cv60, Buy_Rank 1, PullbackBuy_VWAP (L1/MSCAP-1 favoured cell). val_adj +2 (UNDERVALUED MoS +0.0766 x20 = +2, QPass True unhalved), bias 0 (no actionable cells) -> adj 62 vs the 65 floor; required RAW cv is 63. MSCAP-21 leg (i) SATISFIED and more cleanly than 08-06: settled close, gate cleared 2.518% range vs the 3.000% cap (0.48pp margin, vs 0.23pp on 08-06) on SGD 3.92m turnover, after thirteen gate blocks in the prior fourteen sessions. Leg (ii) MISSED by 5 raw. MSCAP-21 is explicit that legs firing on different days is evidence for the champion/challenger shadow test, NOT a near-miss to bridge. MSCAP-22 restatement (its 'the value is already gone / needs raw 65 unaided' framing is PARTLY VOIDED by tonight's regen): valuation help +2 restored, forward view +16.5%, gate PASSED with margin, required raw 63. SHAPE CHANGE: Entry_Price 2.95 vs 2.78 close, Entry_Distance_ATR +1.29 - a buy-STOP 6.1% above market (anchored-VWAP reclaim), not the 2.74 pullback LIMIT of 08-06; even at floor it would have shipped QUEUED and self-limiting. Recorded-not-proposed estimates at f=0.30 (L5, calib_1w null): entry range 2.917-2.983, 1wk 3.07 (+4.1%), 4wk 3.35 (+13.6%), 12wk 3.24 (+9.8%). Note DCF_per_share 31.92 = 11.5x price (38.79 -> 31.92 in one regen) - band is EPV-set (FV 3.01) so the label is sound, but the 'exceeds DCF anchor' guard is silently inert here (systemic #4). RE-ARM: a settled PullbackBuy_VWAP/MomentumContinuation at raw cv >=63 still clearing the gate = a card. A gap-up print tomorrow is NOT it - L6 bars the post-catalyst chase and NVIDIA Q2 FY2027 + US July PCE both price into SGX on 08-27. · news
8AZAztech Global0.57WATCH0.570.55n/a0.63 (+10.5%)0.32 (-43.9%)Aztech looks cheap and deeply oversold, and we still will not touch it. It fails our financial-health screen outright, which blocks new positions regardless of price, and our own twelve-week view sits 44% BELOW where the shares trade today. Tonight's signal reappearing is not recovery - it vanished yesterday only because the shares moved too erratically for us to read them, and returned today because they moved less. Cheap plus broken is not a position. OPERATIONAL: WATCH cv58, ReversalUp_Oversold back from extinction (path -> ->. MSCAP-19 applies: last night's cv0 was GATE SUPPRESSION (3.478% range), not a withdrawn signal, so tonight's 1.754% reappearance is CADENCE, not information. Closed 0.5700 (-0.87%), RSI 17.16 - a new low on the move. Regen improved valuation (UNDERVALUED MoS +0.220 -> +0.2591) but val_adj is only +2: x20 = +5 HALVED on Quality_Pass False. adj 60 misses the floor, but MSCAP-31 governs - report the FIRST binding block in precedence order: the HARD Quality_Pass gate on new entries, not the floor. WATCH is also not a bullish actionable Action. T12 0.32 vs 0.570 = -43.9%. RE-ARM unchanged (SGM-4): leg one satisfied; LEG TWO BINDING AND UNMET - FY2026 revenue stabilisation repairing Quality_Pass, THEN a settled favoured-cell BUY >=65 with a forward view that is not forty percent below price. · news
558UMS Integration2.71WATCH2.712.67n/a2.96 (+9.2%)3.15 (+16.2%)UMS surfaced a signal tonight for the first time in a week, but nothing about the business changed - the shares simply stopped moving enough for our filters to look at them. On our valuation work it is the second-most expensive name in the book relative to what we think it is worth, and it fails the financial-health screen too. The twelve-week view is genuinely attractive at +16%, and it is not enough against a double block. UMS is one of the four names most exposed to NVIDIA's results tomorrow. OPERATIONAL: WATCH cv52, FailedBreakdown_Up_20D - the family L1/MSCAP-1 explicitly discount (~0.25 hit rate, ~60% stop-first). Surfaced ONLY on a 0.05pp gate margin (2.952% range vs the 3.000% cap) on SGD 10.6m; MSCAP-19 - a gate-permitted surface is not new information about the name. Closed 2.710 UNCHANGED, RSI 55.7. SGM-4a double block persists: GROWTH_PRICED_IN MoS -4.3424 -> val_adj -10 (clamped) AND Quality_Pass False. adj 42, twenty-three points below the floor. T12 3.15 vs 2.71 = +16.2%, second-best forward view in the book. Needs a valuation AND quality re-rate, not a chart event. · news
5DDMicro-Mechanics (Holdings)2.79WATCH2.792.77n/a2.91 (+4.3%)2.60 (-6.8%)Micro-Mechanics has now swung across a twenty-point conviction range while its share price moved less than two percent - the reading is not measuring what it claims to on a tape this thin. It is priced above what we think it is worth, our twelve-week view is below today's price, and the signal type is the weakest one we track. We ignore it. OPERATIONAL: WATCH cv48, sixth consecutive carrier session. Instability series extends to NT -> -> -> across closes 2.780/2.780/2.760/2.760/2.810/2.810/2.790 - three double-digit moves at or near flat prices (MSCAP-34: a conviction round-trip on a near-unchanged close is a stability argument AGAINST the name, never progress toward the floor). Closed 2.790 (-0.71%), RSI 51.7. GROWTH_PRICED_IN MoS -1.4343 -> val_adj -7 -> adj 41, twenty-four points short. Blocked four ways: WATCH not actionable; discounted FailedBreakdown family; T12 2.60 vs 2.79 = -6.8%; unstable conviction. Gate passed honestly at 1.792% on SGD 177,444 (1.77x the turnover floor). · news
AWXAEM Holdings9.06WATCH9.068.92n/a10.02 (+10.6%)9.40 (+3.8%)AEM is the book's deepest, most tradeable name and the one most exposed to NVIDIA's results tomorrow - and it is also the most expensive thing we own no part of, trading at many times what our valuation work supports, while failing the financial-health screen. It gave back today's bounce and our twelve-week projection resumed falling after a single flat session. We refused this name twice in August and both refusals are further ahead tonight than they were yesterday. OPERATIONAL: WATCH cv45, first signal in the window, surfacing ONLY because its range fell to 2.870% - a 0.13pp margin under the 3.000% cap - on the deepest tape in the book (SGD 23.9m, turnover leg passing ~239x). MSCAP-19: cadence, not recovery. Closed 9.060 (-2.58%), RSI 48.91 -> 46.22. T12 9.40 = +3.8% but the erosion RESUMED after one flat session (11.38 -> 10.65 -> 9.98 -> 9.96 -> 9.40). GROWTH_PRICED_IN MoS -9.7737 -> val_adj -10 (clamped; eased from -10.844) AND Quality_Pass False = double block no chart event clears; the discounted FailedBreakdown family on top. adj 35. Refusal ledger marked to tonight: 08-12 leg -9.40% avoided (from -7.00%), 08-13 chase leg -17.64% (from -15.45%) - both WIDENED. MSCAP-25: grade both legs every session; MSCAP-12: this never re-litigates the floor. · news
E28Frencken Group2.54WATCHn/an/a2.03Frencken did not trade at all today - an unchanged close on a wide, jumpy range that our liquidity filter rejects. Our own twelve-week projection sits 20% below the current price, so a clean signal here would argue against our own forecast. Nothing to do. OPERATIONAL: NO TRADE cv0. Third consecutive night of a RANGE failure (5.118%, computed from the row's own High/Low/Close) misreported as 'Consensus conflict + low conviction' - the engine.py:710 defect (systemic #2). Closed 2.540 unchanged, RSI 44.4. T12 2.03 = -20.1%, flat for a third session. FAIR_VALUE MoS -1.2266 -> val_adj -6: MSCAP-2 over-penalty FLAGGED, not overridden - treat the FAIR_VALUE band as neutral and lean on the MoS sign, not the x5 magnitude. Quality_Pass True. · news
I07ISDN Holdings0.69WATCHn/an/a0.74ISDN resumed sliding after its August product-launch spike round-tripped in full. It still moves too erratically for our liquidity filter, and it fails our financial-health screen. We waited through the spike and we are not chasing the second leg either. OPERATIONAL: NO TRADE cv0, clean range failure at 3.650% on SGD 3.31m (turnover leg passing comfortably - a volatility rejection, not an illiquidity one). Closed 0.6850 (-1.44%), RSI 51.2 -> 49.4. T12 0.74 = +8.0%. Regen improved it to FAIR_VALUE MoS -0.742 -> -0.5424, val_adj -4 -> -3; Quality_Pass False. The 08-19 Automation Taipei launch was a PR event that fully round-tripped (-12.9%). · news
BN2Valuetronics Holdings1.01WATCHn/an/a0.96Valuetronics finally moved after four unchanged sessions, and its signal promptly vanished - which is the point: last night's reading was an artefact, not a setup. It is priced above what we think it is worth, fails the financial-health screen, and our twelve-week view is below the current price. OPERATIONAL: NO TRADE cv0 (from 52), Pillar_Consensus Indeterminate. BN2 is the GATE CONTROL for the third consecutive night - it passes the liquidity gate honestly (1.980% range on SGD 392,284), so its consensus label is GENUINE, unlike E28/575 (systemic #2). Closed 1.0100 (-0.98%), breaking the four-session unchanged run that produced last night's decisive instability finding (byte-identical OHLC bar, conviction, volume the only differing input). RSI 49.7 -> 45.9. T12 0.96 = -5.0%. GROWTH_PRICED_IN MoS -0.1089 -> val_adj -1; Quality_Pass False (hard gate on new entries). · news
M14InnoTek0.44WATCHn/an/a0.25InnoTek barely traded today - under Schanged hands all session - and it is the only name whose valuation got WORSE in tonight's refreshed work. First-half profit collapsed 64% to SOur twelve-week view is 44% below the current price. There is no case here on any axis. OPERATIONAL: NO TRADE cv0, and now a TURNOVER-leg failure (SGD 43,788, under half the SGD 100,000 floor) as well as a range failure (3.371%) - a clean FAIL_Both on an unchanged 0.4450 close. CORRECTION TO A STANDING CLAIM: the memory has asserted for several sessions that S71 is the ONLY name in this book ever to fail the turnover leg; M14 failed it tonight, so that claim is narrowed. RSI 35.0. T12 0.25 = -43.8%, joint-worst forward view. THE ONLY NAME TO DETERIORATE IN THE REGEN: GROWTH_PRICED_IN MoS -0.330 -> -0.721, val_adj -2 -> -4; DCF_per_share UNAVAILABLE. Quality_Pass False. Refusal ledger: the 08-12 WATCH cv68 refusal marks -12.75% avoided (flat vs last night). · news
575ASTI Holdings0.11WATCHn/an/a0.09ASTI is an eleven-cent stock whose daily swings our filters reject, and our twelve-week view fell further below the current price tonight. The valuation label improved but the practical read did not. Nothing to do. OPERATIONAL: NO TRADE cv0. Closed 0.1140 (-1.72%), RSI 40.5. RANGE failure (3.509%) misreported as consensus conflict for a third night - engine.py:710 (systemic #2). Valuation flipped GROWTH_PRICED_IN -> FAIR_VALUE (MoS -2.3849 -> -1.9014) but STILL clamps to val_adj -10: MSCAP-2 SATURATION for a fourth night - 575 at MoS -1.90 and AWX at MoS -9.77 take the identical haircut despite a 5.1x difference in stated mispricing. T12 0.09 vs 0.114 = -21.1%, deteriorating from -13.8%. Quality_Pass True. · news
S71Sunright0.43WATCHn/an/a0.25Sunright is now the cheapest name in the book on our refreshed valuation work - and it traded under three thousand dollars all day, roughly three percent of the minimum we require to consider a name investable. Our twelve-week view is 42% below the current price. Cheapness has never bought a position here and it does not tonight. OPERATIONAL: NO TRADE cv0, worst performer on the board at -3.37% to 0.4300. Clean FAIL_Both: SGD 2,838 turnover on (under 3% of the SGD 100,000 floor - a third consecutive deterioration, -> 2,838) with a 9.302% range. RSI 34.1. Regen widened it to the largest positive val_adj in the book: UNDERVALUED MoS +0.4594 -> +0.5256, +10 unhalved on Quality_Pass True. T12 0.25 vs 0.430 = -41.9%. SGM-5 in its purest form: a QPass-True UNDERVALUED reading never originates a position absent a bullish chart AND liquidity clearance. MSCAP-2 note: this +10 is worthless on a name trading SGD 2,838/day - the val_adj curve is miscalibrated at BOTH ends for SGX small-caps (systemic #5). · news

Outcome & track record

Accuracy at the time of this record.

0 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.
0.000 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.

This website is for educational purposes only and does not constitute investment advice. Always do your own research and assess your own risk tolerance before making investment decisions.