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 |
|---|---|---|---|---|---|---|---|---|---|
| NVDA | NVIDIA Corporation | — | NO ACTION | 218.36 | 221.79 | 222.43 (+1.9%) | 231.91 (+6.2%) | 235.95 (+8.1%) | Holding Nvidia through a stop breach. The shares closed at 218.36, just below the 221.79 trailing stop we had set, but this was a sector-wide session - nineteen of the twenty names in this mandate fell and the median lost 3.36% - driven by inflation worries and multi-year-high bond yields rather than anything to do with Nvidia. On the same day Jensen Huang reiterated confidence in roughly 70% revenue growth next year, with the Vera Rubin platform now in production shipment and guided to about 20% of data-centre revenue this quarter. The stock still sits above its fifty-, hundred- and two-hundred-day averages and our own twelve-week view is 8% higher. Selling the largest holding into a market-wide down day, on the first close through a self-imposed line, is how you cut a winner. We hold - but we have written down the exact condition that changes our mind. OPERATIONAL: breach -1.55% / -0.51xATR (approx -1.43% adjusting for the ex-dividend that fell on this date). NO origination class fires: L2 needs stop-break AND negative 12wk - 12wk is +8.06%, conjunction FAILS. ADR-0012 arm is PHANTOM ( -3.72%, Gain_ATR -1.25) per learnings advisory (a) - discarded; no tick fires anyway (no 2nd warning code, conviction -5 not >=20, deadview False, zero NO TRADE prints). ADR-0020 erosion 0.00% vs peak_stop 221.79. Label is HOLD, 0.0, no Sell_Rank. USC-5: fade in a still-constructive structure (below EMA5/10/20 at 222.74/222.29/220.09 but above EMA50 214.81 / EMA100 209.03 / EMA200 201.29; Consolidation_Bullish, Trend 64, Momentum 53 - much improved on last night's 26) -> trail, don't trim. G3 forward-view gate blocks the trim. TRAIL HELD at 221.79, deliberately NOT lowered (2.0xATR chandelier = 204.81, which is also the engine's own printed Stop_Price); a trail never ratchets down, so the recorded line now sits above the market and the monitor will keep flagging it. CONVERSION TRIGGER (pre-committed, USC-6/L3): a SECOND consecutive close below 221.79, OR any close below EMA50 214.81, converts to a MANDATORY 30% de-risk at the next open. Ledgered optional:true so the counterfactual grader prices what holding through the breach cost or saved - the optional_trim counterfactual (n=15, median -5.06% 1wk) is the standing argument on the other side. 83% of capital. f=0.30 per L5 (calib_1w -0.226, degenerate). · news |
| ASML | ASML Holding NV | — | HOLD | 1687.43 | 1622.58 | n/a | n/a | 1570.04 (-7.0%) | Holding ASML for one more session, on a technicality we are naming out loud. The chart has deteriorated sharply - the signal engine has gone from a weak bearish reading to no signal at all, the shares have slipped below every short- and medium-term average, and our twelve-week view is now 7% BELOW today's price. It is also the worst position in the book at -7.6%. Set against that, the business news is strongly positive: 2026 guidance of -51.5bn against a ~$45bn consensus, Intel now in high-volume manufacturing of Panther Lake using ASML's High-NA machines, TSMC and Samsung both committed, and EUV orders booked into 2028. That divergence between a strengthening business and a breaking chart is why this is a dated trigger rather than a trim tonight. OPERATIONAL: conviction (-45, clears the ADR-0012 tick (b) >=20 threshold) but tick (b) only fires on an ARMED position and the PROFIT_ARMED flag is PHANTOM ( -7.64%, Gain_ATR -2.67) per learnings advisory (a) - discarded, so no profit-take originates. Note the arm would ALSO have fired tick (c): 12wk 1570.04 vs 1.02xClose 1721.18 = deadview TRUE. Pillar_Consensus flipped Consolidation_Bullish -> Indeterminate; NoTrade_Reason 'Consensus conflict + low conviction'. Close below EMA5/10/20/50 (1709.21/1713.06/1723.77/1725.22), still above EMA100 1663.30 / EMA200 1499.58. L2 FAILS the conjunction: 12wk IS negative (-6.96%) but close is +4.00% ABOVE the stop - no break. Section 9 'sustained NoTrade/Conflicted' FAILS at n=1 - this is the FIRST NO TRADE print (AVOID 64 -> AVOID 45 -> NO TRADE 0); ADR-0012 tick (d) needs 2 consecutive. ADR-0020 erosion -6.56% vs peak 1736.47, inside the 15% threshold. TRAIL HELD at 1622.58 (chandelier 1582.70 is lower; never lower a trail). CONVERSION TRIGGER (pre-committed): a SECOND consecutive NO TRADE / Conflicted print on the 09-11 close satisfies section 9's sustained-NoTrade origination and fires a MANDATORY 30% de-risk at the following open. Materiality stated honestly: position is = 1.04% of book, so a 30% trim is - the discipline matters more than the dollars. Quality_Pass FALSE (FCF yield 0.02%) but that gates new entries only, never forces a sell. · news |
| TSM | Taiwan Semiconductor Manufacturing C | — | HOLD | 428.03 | 415.38 | 434.29 (+1.5%) | 448.88 (+4.9%) | 433.06 (+1.2%) | Holding TSMC, comfortably. It is the strongest chart in the book - above every moving average and down only 1.7% on a day the average name in this mandate fell 3.4% - and the fundamentals are emphatic: August revenue of NT8bn (about 35bn) was up 53.3% on last year and 10.1% on July, with eight-month sales up 39.3%. The shares slipped anyway, which says more about market sentiment than about the business. Nothing here argues for selling. OPERATIONAL: PROFIT_ARMED is PHANTOM ( -2.28%, Gain_ATR -0.96) per learnings advisory (a) - discarded. Tick (c) WOULD have fired had the arm been valid (12wk 433.06 vs 1.02xClose 436.59 = deadview TRUE); even then a 30% trim of rounds to 1 share against a book. Above EMA10/20/50/100/200 (425.63/422.41/419.36/408.32/378.24), marginally below EMA5 429.10; Trend 78, Consolidation_Bullish. Headroom +3.05% / +1.21xATR over the trail. TRAIL HELD at 415.38 (chandelier 407.18 is lower; never lower a trail). ADR-0020 erosion 0.00% vs peak 415.38. Quality_Pass TRUE, F_Score 9, Quality_Rank 4/20. TSM ex-div on 2026-09-16; FOMC 2026-09-16. 12wk is only +1.17% - weak forward view, logged under systemic finding 4. · news |
| — | Advanced Micro Devices, Inc | — | NO ACTION | — | — | n/a | n/a | n/a | Passing on AMD, and it is the closest call of the night. The company's news is genuinely good - at Citi's TMT conference the finance chief raised the AI-chip market ceiling to trillion by 2030, server-CPU revenue is guided up more than 80% in the second half, the MI450 starts shipping at the end of this quarter, and a fresh Buy initiation carries a target, some 19% higher. What stops us is not the business but the price and the setup: this is a signal family in which every trade this book has ever taken has stopped out within a week, and the model's own twelve-week projection offers under 1% above today's price. A good business at a bad entry, seven points short of our threshold. OPERATIONAL: adj = 68 + 0 (val_adj) - 10 (USC-3 floor) = 58, SEVEN short of the 65 floor. Quality_Pass is TRUE (F_Score 8, Quality_Rank 6/20) - the hard gate is NOT the blocker here. val_adj SUPPRESSED to 0 per note B: DCF 171.74 and FV/sh 42.64 against a 503.60 price is a broken anchor, not a -11.22 valuation signal. scorecard_bias -10 per USC-3's PRE-COMMITMENT (the scorer prints -2 at n=5 but hit_4w is 0.00 and this book's realized record in FailedBreakdown_Up_20D|buy is 3/3 stop-first, median_err_1w -9.77%) - systemic finding 5, and tonight is its strongest evidence: -2 would have produced adj 66 and a filled buy on the day the complex fell 3.36%. Entry 503.60 = close (Entry_Distance_ATR -0.0, range 498.46-508.74), stop 493.33, target 544.68 -> R:R 4.00:1. 12wk 507.88 = +0.85% - WARNING exceeds DCF anchor 171.74, projection assumes growth beyond priced-in. 1wk 515.92 (+2.45%, f=0.30 per L5). Anticipate slot (AVOID 67 -> BUY-ANTIC 76 -> BUY 68). AMD fell -3.36% this session and its 530.63 buy-stop from 09-09 never filled (day high 516.33). |
| — | ASE Technology Holding | — | NO ACTION | — | — | n/a | n/a | n/a | Declining ASE Technology for a fourth session. The chart has been one of the strongest in the mandate, but the company is burning cash hard - free cash flow is negative to the tune of 14% of its market value - as it funds a large capacity build, and it fails our balance-sheet screen outright. Our valuation work also has it priced well beyond any defensible anchor. A strong chart on a cash-burning balance sheet, in a signal family that has stopped this book out every time, is not a trade we take. OPERATIONAL: BLOCKED twice over. (i) Quality_Pass == FALSE - FCF_Yield -14.2%, F_Score 6, Quality_Rank 16/20; HARD gate on new entries, substantiated by the real $2bn LEAP capex build, not an artifact. (ii) adj = 68 - 10 (val_adj, MoS -10.85) - 10 (USC-3 floor) = 48, SEVENTEEN short of the 65 floor. Anticipate precondition IS still met on a trailing basis (BUY-ANTIC on 09-08 and 09-09 = 2 of last 3, conviction 68 >= 55 floor) but is blocked by the quality gate AND by a second independent block: the live row prints Entry_Price 39.86 = its own close with Entry_Distance_ATR -0.0, i.e. there is NO buy-stop above the market to queue - carrying forward the 09-09 trigger of 40.59 would violate the action-price provenance rule. ASX is one of this book's three realized FailedBreakdown stop-outs (07-10, -8.65%). Fell -3.25% this session; the 40.59 buy-stop never filled (day high 40.30). Note DCF_per_share is BLANK and FV/sh is 3.48 against a 39.86 price - a case for note B suppression too, which would give adj 58, still ten short. Either way: NO ACTION. f=0.30 per L5. |
| — | Lumentum Holdings | — | NO ACTION | — | — | n/a | n/a | n/a | Still not buying Lumentum - and this session settled the argument. For three sessions it carried the strongest buy setup on the board and was blocked only by our balance-sheet screen, which reads a 5-out-of-9 financial-strength score and free cash flow of barely a third of a percent. It has now fallen 5.4% in a single day, the setup has decayed from a top-conviction anticipation signal to a simple watch, and the entry trigger we would have used was never touched. The screen we were told was being over-strict was, on this occasion, exactly right. OPERATIONAL: BLOCKED on two independent grounds. (i) Quality_Pass == FALSE - F_Score 5, Quality_Rank 14/20, FCF_Yield 0.35%; HARD gate on new entries; USC-3 forbids a gate override to admit a chase. (ii) NO LIVE TRIGGER: the row now prints Action WATCH cv61 with Entry_Price 935.70 = its own close and Entry_Distance_ATR -0.0 - there is no self-limiting buy-stop above the market to queue, and carrying forward the 09-09 trigger of 1006.49 would violate the action-price provenance rule. The ADR-0012 anticipate precondition IS still met on a trailing basis (BUY-ANTIC 100 on 09-08, 93 on 09-09 = 2 of last 3; conviction 61 >= the 55 floor) - the count was never the blocker here. VALIDATION: the 1006.49 buy-stop would NOT have filled (day O/H/L/C 959.43 / 994.50 / 933.82 / 935.70) - USC-4's reclaim-gated buy-stop committed zero capital, re-validated. Valuation anchor remains unstable (EPV -625.57, MoS swung -34.07 -> -7.75 on a single 7-day refresh) so val_adj is suppressed to 0 under note B - systemic finding 6. f=0.30 per L5. |
| — | Micron Technology, Inc. | — | NO ACTION | — | — | n/a | n/a | n/a | The Micron re-arm we flagged last night did not happen, and waiting saved money. We wrote that a second anticipation signal on this close would clear Micron's last hurdle and make it the book's next buy. Instead the signal weakened to a simple watch and the shares fell 4.9% in a session where memory names led the sector lower. The buy-stop we would have used at 1,045.66 was never reached - the stock opened at 996.78 and never traded above 999. The quality of the business is not in question; the entry simply was not there. No trade. OPERATIONAL: the 09-09 prediction ('RE-ARM LIVE: a second BUY - ANTICIPATE print on the 09-10 close makes it, and MU has no gate left to fail') is now DEAD, not deferred: MU printed WATCH 61 (trail BUY 76 -> BUY 68 -> BUY-ANTIC 83 -> WATCH 61 =). adj = 61 - 7 (val_adj, MoS -1.38) = 54, eleven short of the 65 floor, and the row is no longer in the buy pool at all (WATCH, not a bullish actionable Action). Quality_Pass remains TRUE and best-in-class among the anticipate names (F_Score 8, Quality_Rank 2/20, FCF 2bn, D/E 0.06) - the gate was never MU's problem. News classified OVERRIDES: memory-cost and AI-spend concerns drove the complex lower on this date, which is a sector-level demotion of the setup rather than a company-specific one. MU FQ4 reports 2026-09-30; management has already flagged a moderation in the rate of price increases, so L6 (no post-earnings gap-chase) will govern the next look. Ledgered so the counterfactual grader prices what one more session of waiting was worth - this time it was worth +4.90%. f=0.30 per L5. |
Outcome & track record
Accuracy at the time of this record.
45 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 →
-0.226 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.
53 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.