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 |
|---|---|---|---|---|---|---|---|---|---|
| ASML | ASML Holding NV | 1714.88 | SELL ALL | 1714.88 | 1611.58 | n/a | n/a | 1560.64 (-9.0%) | Exit the position in full. ASML rose 4.2% with the whole chip sector on Friday, and the model's sell case got STRONGER rather than weaker through the rally - it now carries the most emphatic sell reading on the board, ranked first among everything the book could dispose of, with the engine's own instruction reading 'target full exit'. The shares still sit below their 20- and 50-day averages even after the bounce, and the model's twelve-week view of 1,560.64 is 9% BELOW Friday's close. The business remains genuinely excellent - 2026 revenue guided to 43-45bn euros at 54-56% gross margin, a growing high-margin service annuity, and a Strong Buy consensus with a target near - but none of that is new information, and the next real catalyst (Q3 results) is five weeks past this trade's horizon. When a strong company's price action has decisively broken and the model's own forward view has gone negative, the position is dead money at best. Sell the six shares and take the 6.1% loss. Operational: EIGHTH re-fire of this exit; L2 stop-break+dead-12wk basis, NOT a profit-take - the PROFIT_ARMED flag on this row is the known defective T12 disjunct (position is 6.14% underwater, Gain_ATR -2.17) and is discarded. prints 77.89% but Position_Status CLOSING + 'target full exit' resolves to 100% of. Proceeds; realized -. G1/G2/G3 do not bind - this is the skill's stated contrast case (trend-based exit on a losing, bearish, negative-12wk position). Scorecard cell FailedBreakout_Down_20D|sell is n=1 report-only, bias 0, no adjustment applied. 1wk/4wk n/a - a CLOSING row prints no Target_Price; model next-session band 1,641.32-1,759.90. Execute at the Tue 2026-09-08 open (Mon 09-07 is Labor Day). · news |
| TSM | Taiwan Semiconductor Manufacturing C | 428.91 | HOLD | 428.91 | 409.63 | 434.69 (+1.3%) | 448.19 (+4.5%) | 416.00 (-3.0%) | Keep all three shares and raise the protective stop. After three sessions in which the model could find no tradeable setup here at all, TSMC came back to life on Friday: the signal flipped to a strong-buy pattern at high conviction, the pillars realigned, and the shares reclaimed their 20-, 50- and 200-day averages in one move. The business case never wavered - July revenue up 44.7% on the year with AI demand called 'extremely robust', record 2026 capital spending of -64bn, and a $265bn Arizona build-out - and now the price action has caught up with it. The four-session-old plan to trim one share is therefore cancelled outright rather than rolled forward: the reason for it no longer exists. Raise the stop to 409.63 and let the position run. Operational: trail raised 405.91 -> 409.63 (live Stop_Price), 4.50% / 2.00xATR headroom, fresh ADR-0020 watermark. Valuation layer SUPPRESSED on this name by G2 - the DCF of 46.22 is the known ADR-ratio artifact (TSM ADR = 5 ordinary shares; per-ADR anchor should be ~5x,), so MoS_FV -13.82 is unusable and val_adj forced to 0. Estimates at f=0.30 (calib_1w -0.226 degenerate, L5). Ex-div on 2026-09-16, 6 sessions past execution, outside the <=5-day catalyst window. · news |
| NVDA | NVIDIA Corporation | 230.36 | HOLD | 230.36 | 215.95 | 244.37 (+6.1%) | 277.06 (+20.3%) | 256.92 (+11.5%) | Hold all and raise the trailing stop. Nvidia is within 2% of an all-time high and the model's conviction has printed 93 for a second straight session - the highest reading this book has ever recorded on any name. The news flow behind it is substantial and fresh: the 9bn acquisition of Hugging Face closed Thursday, bringing in a platform used by 18 million developers and 200,000 companies; a 5bn investment in MediaTek was announced Monday; and reports of an imminent Anthropic IPO are re-rating Nvidia's $10bn stake. The one caution on the chart - momentum not quite keeping pace with price - is the sort of thing that usually resolves upward in a trend this strong, so the right response is a tighter stop, not a smaller position. Operational: G1 governs - RSI_DIV on a bullish high-conviction trend is a trail-raise, never a trim. NOT armed (Gain_ATR 0.47 << 3.0; T12_Progress_Pct 89.66, fractionally under the 90 threshold - one session from arming, watch it), so ADR-0012 never engages. Trail RAISED 215.33 -> 215.95 on the live chandelier_trail; 6.26% / 1.89xATR headroom; fresh ADR-0020 watermark. Estimates at f=0.30 (L5). 12wk 256.92 EXCEEDS the 163.98 DCF anchor - projection assumes growth beyond priced-in. Ex-div on 2026-09-10, two sessions after execution, no action. This is 83.5% of equity. · news |
| AVGO | Broadcom Inc | 357.89 | WATCH | 357.89 | 332.81 | n/a | 373.55 (+4.4%) | 281.48 (-21.4%) | Not buying. Broadcom is the only large chip name that failed to join Friday's sector-wide surge, closing up just 0.2% on a day the median semiconductor stock rose 4.3% - a direct verdict on its latest results, where third-quarter AI revenue of 7bn (up 221% on the year) was overshadowed by a fourth-quarter forecast of 8bn that fell short of expectations. The model does flag it as oversold and worth a bounce, but the trade it proposes offers roughly of upside against of downside, which is not a bet worth taking at any level of confidence. A stock the whole market just passed over is cheap for a reason. Operational: adj 65 clears the floor EXACTLY (75 val_adj -10, bias 0) then fails on three independent grounds - (i) R:R 0.62:1 on the model's own levels (+15.66 reward vs -25.08 risk), which CLAUDE.md section 9 makes the binding constraint on return-seeking; (ii) ALL THREE pillars bearish-context (Consolidation_Bearish, Trend 36 / Momentum 37 / Volatility 38) with the close below EMA20 373.55, EMA50 382.84 AND EMA200 370.15 - the cv75 is reversal-pattern scoring, not alignment; (iii) single print after three consecutive AVOID sessions (68/71/62/75) in a family with ZERO graded buy samples in this book. Re-arms only on pillar realignment plus a target that pays its own risk. · news |
| MU | Micron Technology, Inc. | 1016.59 | WATCH | 1016.59 | 969.00 | n/a | 1107.13 (+8.9%) | 1134.12 (+11.6%) | Not buying, for a third session - and this is the closest call on the board. Micron jumped 6.1% on Friday as the memory sector re-rated on Dell's $95bn AI-server order backlog and continued tightness in DRAM and NAND pricing, and the setup itself is the best-shaped thing this book has seen in weeks: trading above every major moving average, ranked first among buy candidates, with a target offering nearly twice the downside risk and a twelve-week view 11.6% above today's price. What stops it is history, not shape. Every buy this book has taken in this particular pattern has now been graded, and not one of them reached its four-week target - buying a 6% one-day jump in the one setup that has never worked here is a mistake this book has already paid to learn. Waiting for the pullback. Operational: adj 60 = 76 - 6 (val_adj, MoS -1.16, QP pass) - 10 (USC-3). The FailedBreakdown_Up_20D|buy cell CROSSED n=5 into actionable this week at hit_1w 0.40 / hit_4w 0.00; the scorer computes -2 but USC-3 pre-committed this book to the -10 floor at exactly this moment - see systemic finding 5, the disagreement should be settled at the next learnings review since -2 would have admitted this card. Compounded by L6 (gap-chase) and L1. Size modulator would have been 0.75x. RE-ARMS on a PullbackBuy_VWAP print or a pullback toward EMA20 944.38 with the family intact. FQ4 earnings 2026-09-30. · news |
| SNDK | Sandisk Corporation | 1740.00 | WATCH | 1740.00 | 1581.00 | n/a | 1964.76 (+12.9%) | 1869.50 (+7.4%) | Not buying. Sandisk was the single best performer on the board, up 11.9% in one session on the memory-pricing re-rating - it opened at 1,585 and closed at 1,740, right on its high of the day, after three straight sessions in which the model rated it no better than a watch. That is the textbook shape of a chase: the entry is at the top of a vertical move and the protective stop sits 9.1% lower, back at the session's opening low. Paying up after an 11.9% gap and accepting a 9% stop for a target that offers 1.4 times that risk is not a trade this book should take, however good the memory cycle looks. Operational: adj 60 = 76 - 6 (val_adj, MoS -1.11, QP pass) - 10 (USC-3, same actionable-cell hardening as MU). Closed ON its high; stop 1,581.00 = -9.14% / 1.41xATR back to the session low; R:R 1.41:1 does not pay for that. L1 + L6 both bind. Size modulator would have been 0.75x. RE-ARMS on a pullback toward EMA20 1,534.87 or a PullbackBuy_VWAP print. · news |
| TSLA | Tesla Inc | 354.08 | WATCH | — | — | n/a | n/a | 324.83 | Not buying, and the decision not to queue this last night was validated within a single session. Tesla was the ONLY decliner on a board where names rose, giving back 5.9% and erasing the entire Cybercab-driven pop that had briefly triggered a buy signal on Thursday. The model has gone silent on it again. Operational: BreakoutUp_20D cv74 (adj 64, quality-gate blocked) on 09-03 -> NO TRADE cv0 on 09-04, with the close reversing 376.36 -> 354.08. FOURTH clean validation of US-1 (one-session breakout-ADD that fades by the next open) and of the no-chase queue discipline. NoTrade_Reason verbatim: 'Consensus conflict + low conviction'. Logged as counterfactual evidence FOR the current filter, against systemic finding 5's argument that the book over-filters. |
| LITE | Lumentum Holdings | 881.26 | WATCH | 881.26 | 841.04 | n/a | 989.37 (+12.3%) | 956.71 (+8.6%) | Not buying. Lumentum rose 4% with the sector but the model still rates it only a watch, well short of the confidence needed to commit capital, and it fails the quality screen on fundamentals - which blocks a new position outright regardless of the chart. Operational: adj 47 = 57 - 10 (val_adj clamped, MoS -34.07 - the most extreme reading in the book). Quality_Pass FALSE = hard block on new entries (F-Score/FCF screen failed). 18 points short of the 65 floor even before the gate. Fourth consecutive session blocked on the same two grounds. |
Outcome & track record
Accuracy at the time of this record.
44 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.