SG Mid/Small Cap — End-of-Day Verdict
NO ACTION (both sides) - 26th written EOD verdict since 07-03; the book printed its FIRST BUY of the standalone era (AWX cv69, Buy_Rank 1) and refused it on the quality gate, the conviction floor, the weak FailedBreakdown cell and a 0.36-ATR stop - the first session where the blocking control was quality, not liquidity
Recommendations
0 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 |
|---|---|---|---|---|---|---|---|---|---|
| AWX | AEM Holdings | 9.63 | NO ACTION | 9.63 | 9.46 | n/a | 10.58 (+9.9%) | 9.93 (+3.1%) | AEM finally gave us a buy signal after thirteen sessions of silence, and we are still not buying it - but for a better reason than before. The trading screen that had been rejecting it cleared tonight; what stops us now is our quality test, which AEM fails on profitability measured through March 2026. That is the problem: those figures predate the turnaround the company is actually delivering, with first-quarter revenue up 36% and margins tripling, and analysts have just lifted their target price 77% to S. We would rather wait one fundamentals update for the quality screen to catch up than buy on a weak signal type with a stop so tight that an ordinary day's trading would likely trigger it. OPERATIONAL: gate PASSED (2.70% range, SGD 36.89m turnover), Buy_Rank 1, Trend 77 / Mom 52 / Vol 45, RSI 57.6, T12 9.93 = +3.1% (first positive forward view here). Blocked by (1) Quality_Pass False - HARD new-entry block, independent of val_adj; (2) adj 59 < 65 floor; (3) L1/MSCAP-1 FailedBreakdown cell (~0.25 hit, ~60% stop-first); (4) stop 9.46 = 0.36 ATR, sitting on today's low. NOTE: under MSCAP-2's neutral-band reading (val_adj 0) this fuses to 69 and would CLEAR the floor - the verdict holds only because the quality gate is a separate hard block. MSCAP-2 root cause diagnosed: FV_method=EPV picked 0.810/sh while DCF_per_share = 10.269 (6.6% ABOVE close) - a 12.7x anchor disagreement, on TTM AsOf 2026-03-31, pre-turnaround. Ledgered optional:true with action_price 9.63 so the counterfactual grader prices this refusal (partial fix for finding 6c). Re-arm: Quality_Pass repair + settled FAVOURED-cell BUY >= 65 + stop wider than ~1 ATR. · news |
| PCT | PC Partner Group | 3.16 | WATCH | — | — | n/a | n/a | 4.47 | PC Partner rose 6.4% today, carries by far the best twelve-week outlook in this book at about +41%, trades near fair value, and reports first-half earnings likely to double on better pricing and margins. We still could not buy it, for the fourth time in five sessions, because a screen that rejects stocks with wide daily trading ranges caught it - and the range was wide precisely because the stock jumped. The entry level it last offered us is now 15% below where it trades. We are logging this as a flaw in our own screen rather than overriding it mid-flight, because bending a risk rule to justify a trade we want is how books get hurt. OPERATIONAL: Trend pillar (book high), RSI 65.7, close 2.970 -> 3.160, turnover SGD 11.32m (2.6x), T12 4.47 = +41.5%. Fail_Spread on a 7.91% range (2.93-3.18) vs the 3.00% cap; range_to_atr 1.55 would pass both the US (2.0) and HK (3.0) rules. Never reached conviction scoring. IMPORTANT DISTINCTION: the 08-06 refusal at the 2.74 limit was a FLOOR refusal (cv62 < 65), not a gate refusal - now +15.3% adverse; separately the QUALITY-gate refusal of 8AZ at 0.62 is -2.4% and vindicated. Opposite track records - see systemic finding 1. No action_price recorded: the gate suppressed Entry_Price and the provenance rule forbids inventing a limit the live row does not print, which is exactly why gate-blocked names stay ungradeable (finding 6c residual). Re-arm: settled PullbackBuy_VWAP/MomentumContinuation BUY that clears the gate AND prints raw cv >= 65. DO NOT tune the 3% gate - MSCAP-1 / SGM-3. · news |
| BN2 | Valuetronics Holdings | 1.00 | WATCH | — | — | n/a | n/a | 0.91 | The corrupted price history that made Valuetronics unreadable for us has been fixed at the data provider, so we can assess this stock properly again. Its indicators had been showing a heavily overbought stock when it was simply trading flat; they now line up with the actual tape. Nothing about that changes our view of the business - the shares still generate no buy signal and the company still fails our quality screen - but we are no longer flying blind on it. One residual caution: the provider's dividend record is still wrong, so we will re-check that the numbers hold before relying on them fully. OPERATIONAL: SEVERITY-1 RESOLVED AT SOURCE. Ladder re-based in ONE bar - EMA200 0.6944 -> 0.9216 (+32.7%; alpha = 2/201 = 0.995% is the max from new data, so the history was re-fetched, not absorbed), EMA20 0.8067 -> 1.0154, EMA50 0.7822 -> 1.0183, VWAP 0.7920 -> 1.0177, VWAP_Distance +24.99 -> -1.74, RSI 79.86 -> 44.02, ATR 0.0585 -> 0.0266, T12 1.21 -> 0.91. The 08-11 re-arm leg 1 (ladder must re-converge with the raw series) is SATISFIED -> UNMUTED from this entry forward. CAVEAT: confirm the ladder holds on the 08-13 close - a single-bar re-base is also what a bad adjustment looks like. No trading case: NO TRADE cv0, Indeterminate consensus, Quality_Pass False, T12 -9.0%. Yahoo STILL publishes the bogus 'SGD 0.30 dividend, ex 2026-08-07' (= 0.2857 x 1.05) against an actual ~SGD 0.06 drop, so the vendor's price series is fixed while its dividend field is not. Systemic finding 2 downgraded to RESOLVED-AT-SOURCE; the fetch.py quarantine fix STAYS high-priority - nothing in our pipeline caught this across five sessions. · news |
| M14 | InnoTek | 0.51 | WATCH | 0.51 | 0.47 | n/a | 0.53 (+3.9%) | 0.29 (-43.1%) | InnoTek stopped falling and flipped to a tentative recovery reading tonight, its first non-negative signal in weeks. It is not a buy: the reading is only a watch-level flag, the stock still fails our quality screen, the broader trend remains down, and our own twelve-week model still sees it more than 40% lower. A single stabilising day in a downtrend is not an entry. OPERATIONAL: FailedBreakout_Down_20D AVOID cv56 -> ReversalUp_Oversold WATCH cv68; entry 0.51 / stop 0.47 / target 0.53; +2.0% to 0.510, RSI 34.0 -> 38.0, gate passed on a 2.94% range. WATCH is NOT in the buy pool (STRONG BUY / STRONG BUY - ADD / BUY / BUY - ADD only). Even as a BUY it fails: val_adj -7 -> adj 61 < 65, Quality_Pass False (hard new-entry block), consensus still Consolidation_Bearish, Trend 36, T12 0.29 = -43.1% (worst in book). SGM-4b, 10th instance - now in its milder WATCH form. No action_price ledgered as optional: a WATCH label is not a declined entry. |
| 8AZ | Aztech Global | 0.61 | WATCH | 0.61 | 0.61 | n/a | n/a | 0.34 (-44.3%) | The Aztech entry we declined on quality grounds six sessions ago is down 2.4%, and the stock has now turned outright bearish - so that refusal was right. It screens as cheap, but it is cheap because the price keeps falling, and our twelve-week model sees it more than 40% lower still. Cheapness on its own has never been a reason for us to buy anything. OPERATIONAL: the cv80 ReversalUp_Oversold declined 08-06 at a 0.62 entry (on Quality_Pass False, NOT on conviction - raw 80, val_adj +2, adj 82, comfortably over the floor) closed 0.605 tonight = -2.4%. Tonight: FailedBreakout_Down_20D cv67 AVOID, consensus flipped Consolidation_Bullish -> Consolidation_Bearish, Trend, RSI 19.7, T12 0.34 = -43.8%. This is the QUALITY-gate leg of systemic finding 1 and it is vindicated, in contrast to the conviction floor (PCT +15.3% adverse). Val_Signal improved to UNDERVALUED MoS +0.220 purely because the price fell - textbook SGM-5. Re-arm: FY2026 revenue stabilisation repairing Quality_Pass FIRST, then a settled favoured-cell BUY >= 65. |
| 575 | ASTI Holdings | 0.14 | WATCH | — | — | n/a | n/a | 0.14 | A micro-cap in this book jumped 12% today on eight times its normal trading volume, with no news we could find to explain it. Our screen refused it on the width of the move, and on this occasion we think the screen was right: an unexplained double-digit move in a stock our valuation work puts at a fraction of its market price is speculation, not an opportunity. We are recording this alongside our complaints about the same screen blocking better names, so the case stays honest. OPERATIONAL: 0.123 -> 0.138 (+12.2%) on 36.6m shares / SGD 5.05m turnover vs a typical SGD 0.65m; 13.77% intraday range (0.121-0.140), the widest in the book -> unambiguous Fail_Spread (range_to_atr 2.27: would fail the US 2.0 rule, pass HK's 3.0). Trend pillar, RSI 51.0 -> 64.0, T12 0.14 = +1.4%. News NONE FOUND: bounded search surfaced nothing dated near today; most recent item is May 2026 (CEO Eddie Ng acquired 40m shares at 9c to become controlling shareholder). EPV FV 0.036/sh vs a 0.138 close drives MoS -2.440 -> val_adj -10 (EPV 0.036 vs DCF 0.107 = 3.0x anchor disagreement, same MSCAP-2 mechanism as AWX). Counter-exhibit for systemic finding 1: the spread gate working as designed. · news |
| S71 | Sunright | 0.49 | WATCH | — | — | n/a | n/a | 0.28 | Sunright remains the cheapest name in the book on our valuation work and we still will not touch it, because it trades too little. Barely Schanged hands today against a Sminimum we require, and our twelve-week model sees the shares more than 40% lower. A stock we could not exit in a hurry is not a position at any price, however cheap it looks. OPERATIONAL: flat at 0.490, Fail_Both - turnover SGD 43,512 (~56% below the SGD 100k floor, the book's ONLY genuine turnover failure tonight) and a 4.08% range. Largest val_adj in the book at +9 (UNDERVALUED, MoS +0.459, QPass True), which would lower the required raw conviction to 56 - still far above anything the chart prints (Trend 51, T12 0.28 = -42.9%). SGM-5 stands. Needs genuine turnover AND trend repair, in that order. |
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.