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 |
|---|---|---|---|---|---|---|---|---|---|
| TSM | Taiwan Semiconductor Manufacturing C | 416.00 | SELL 30% | 416.00 | 403.96 | 423.05 (+1.7%) | 439.50 (+5.6%) | 399.51 (-4.0%) | We are taking a further 30% off TSM because the model's own 12-week view has now sat below the share price for five straight sessions - 399.51 against a 416.00 close, about 4% under where it trades today. Almost everything else about the position improved tonight: the stock rose 0.95%, conviction climbed from, it reclaimed its 100-day average and it never came near the line we said in advance would take the rest. The business is fine - July revenue rose 44.7% year over year and 3nm is ramping into strong demand - but the Street's consensus and the crowd's fair value disagree by 42% on the same stock, and we would rather hold a smaller piece while that argues itself out. ADR-0012: armed (T12_Progress 104.13) + tick (c) ONLY, now -3.96% through the 1.02x dead band of 424.32. Tick (b) moved the WRONG way (conviction +14) and there is no second Exit_Warning code - verified against exit_warnings.py, not hand-derived. CORRECTION TO 08-19: that card's escalation to profit_take_2 rested on a hand-derived EMA10_BREAK defined as 'two consecutive closes below EMA10'; the engine's actual definition (_detect_ema10_break) requires Close<EMA10 AFTER RSI>70 for >=5 prior bars, and TSM's RSI topped at 56.7 across the whole window. The Tier-2 leg was invalid, the correct 08-19 tier was 30%, and that derivation is retired - tonight is the 30% base with nothing to escalate it. G3 does not block (negative 12wk IS its trigger). G2 suppresses the MoS -14.46 amplify for a 5th session: DCF 45.97 vs 416.00 = 9.05x broken anchor, val_adj 0. Bias +10 from FailedBreakdown_Up_20D|sell (n=6, hit_1w 1.000) -> adj 66. Stop raised 403.69 -> 403.96 (EMA100); model chandelier 392.50 is looser; the 403.85 hard line is RETIRED as duplicative (11 cents apart). Retained risk on :. Realized on vs the 438.02 basis: -. MATERIALITY: on a stub = 0.13% of book - the rule fires so the card is written (ADR-0012 bans prose-only trims; USC-6/L3; optional_trim counterfactual n=13 median -5.14%/-15.33% verdict cost), but see systemic finding 6 - the framework has no minimum-trade-size floor. · news |
| ASML | ASML Holding NV | 1750.31 | HOLD | 1750.31 | 1734.44 | 1785.98 (+2.0%) | 1869.36 (+6.8%) | 1829.85 (+4.5%) | Holding all and nudging the stop up. ASML steadied after two sharp down days, closing flat and roughly 1% above the risk line that forced last night's trim, and it still sits well above its 50-, 100- and 200-day averages. The sell-off traces to reports that a Chinese state-backed firm has started producing its own immersion DUV lithography machines - about 5 this year and 20 next, against the 131 ASML shipped last year. Bank of America puts the worst case at EUR 1.4B of 2027 sales, or 2.4% of the group, and calls the reaction an overreaction; 40 of the 44 analysts covering the stock still rate it Buy or Strong Buy. A 2.4% revenue risk does not justify a 7% de-rate, so we keep the core. ADR-0012: armed (T12_Progress 95.65) with ZERO ticks - conviction -7 vs the 20-pt bar, 12wk 1829.85 alive at +4.54% over the 1785.32 band, no second code (verified against exit_warnings.py: RSI_DIV needs a 20-bar price high, CLIMAX_VOL and EMA10_BREAK both need RSI>70 vs 49.2 actual, AVWAP_BREAK contradicted by VWAP_Distance +2.48%), no NO-TRADE run. Armed-without-a-tick = MANDATORY TRAIL-RAISE and it is legal tonight: EMA50 ticked 1733.79 -> 1734.44. STOP RAISED - 65 cents, taken because a mandatory raise is taken whenever legal (same discipline as last night's 27-cent NVDA raise). Model chandelier 1631.26 is far looser. STANDING OBJECTION: conviction has gone -> over four sessions, a 34-pt collapse, but tick (b) reads vs the PRIOR EOD only and each step is -20/-7/-7, so nothing fires; stated, not acted on, consistent with the last two entries. Marks - / -4.20%; risk to the new line. Not a buy either: val_adj -10 on MoS -4.92 takes cv47 to adj 37. Re-arm for an ADD: raw conviction >= 75 on a bullish-actionable print. · news |
| NVDA | NVIDIA Corporation | 216.85 | HOLD | 216.85 | 206.32 | 220.05 (+1.5%) | 227.52 (+4.9%) | 236.63 (+9.1%) | Holding all into next week's results and leaving the stop where it is. Nvidia was flat on a day the chip complex bounced, and it still sits above its 20-, 50-, 100- and 200-day averages with a 12-week view about 9% above today's price. Q2 lands 2026-08-26, four sessions away, and it is the single biggest scheduled event on this book: the Street expects roughly of revenue and of earnings, Bank of America models -95B - a -4B beat on the company's own guide - and frames it as the possible start of a multi-quarter upgrade cycle. At about 25x forward earnings against a ten-year average near 62, the setup is not a valuation stretch. The discipline into a print like that is to hold and trail, never to pre-trim. ADR-0012: armed (T12_Progress 91.64) with ZERO ticks - conviction -7, 12wk 236.63 alive at +9.12% over the 221.19 band, no second code (verified in source), no NO-TRADE run. Armed-without-a-tick mandates a raise WHERE ONE IS LEGAL and tonight none is: the 2.0x-ATR chandelier prints 206.18, BELOW the standing 206.32, and EMA100 205.33 is lower still. A trail never moves down, so the stop HOLDS at 206.32 - recorded explicitly so the non-raise is not misread as drift. Catalyst-aware hold ACTIVE (horizon band +/-6.75%): trim only on a confirmed lower high or a decisive stop break. Only EMA10 218.90 overhead, by 0.94%. Marks - / -4.39%; risk to stop = 0.27% of book. NVDA is 73.8% of capital - a signal-defined position, not a shape objection. Not a buy: val_adj -10 on MoS -2.56 takes cv54 to adj 44, and the row prints HOLD. L6: no queue on this book may live into 08-26. · news |
| MU | Micron Technology, Inc. | 974.33 | WATCH | 974.33 | 929.07 | 1009.45 (+3.6%) | 1091.41 (+12.0%) | 1080.36 (+10.9%) | Not buying Micron tonight despite the first genuine buy signal this book has seen in three sessions. The news could hardly be better - the stock is up 207% this year, has crossed a trillion valuation, its high-bandwidth memory revenue grew 245% year over year and its entire 2026 HBM capacity is already sold out, with UBS lifting its target to from That is also the problem. The same coverage notes earnings estimates have risen just 1.2% in a month, so the multiple is being carried by price rather than upgrades, and the specific setup here - a failed-breakdown entry on a name that just closed up 4% - is the exact pattern that has lost this book money three times out of three. We want a pullback entry or a materially stronger signal before committing capital. ARITHMETIC: cv69, quality gate PASS, Entry_Distance_ATR 0.00 (not extended, cleanly at the model entry), Buy_Rank 1. val_adj -7 from MoS -1.35 x 5 - and this one is NOT G2-suppressible: DCF 772.94 vs 974.33 = 1.26x, the cleanest anchor on the board after ASML, so note B's broken-anchor carve-out does not apply. FailedBreakdown_Up_20D|buy has now reached n=5 and turned ACTIONABLE at a printed -2, which fires USC-3's pre-committed n>=5 trigger for the FIRST TIME: read it at the -10 floor on this book's 3/3 stop-first record (hit_1w 0.0, median_err_1w -9.77%). 69 - 7 - 10 = adj 52, 13 under the floor. At the printed -2 it is 60, still 5 short - it fails under BOTH readings, so the hardening deepened a rejection rather than manufacturing one. L1 discounts the family before any of this. Size modulator would have been 0.75x (MoS in (-3,-1]). RE-ARM: raw conviction >= 82 on a bullish-actionable print, OR a PullbackBuy_VWAP confirmation, which USC-3 accepts in place of the full-size restriction. The 47-day-stale overlay's MU override is noted and is not a licence - and an overlay cannot open a conviction floor. · news |
| MRVL | Marvell Technology | 251.01 | WATCH | — | — | n/a | n/a | n/a | Watching, not buying. Marvell was the strongest name on the board for a second straight session - up 5.8% today and 15.9% over two, above every moving average we track - but it fails our quality screen and the signal engine declines to name a setup at all, so there is nothing to act on. Results land 2026-08-27. ENGINE OBJECTION (2nd session, systemic finding 5): prints AVOID / NoSignal / cv0 with Exit_Strategy 'none' and no entry, stop, target or rank, while Trend reads Bullish and Pillar_Consensus reads Consolidation_Bullish above EMA10 225.95 through EMA200 163.60 on 28.7M shares. Whatever the cascade's reason for declining a family, the row should surface as a WATCH with a re-arm trigger rather than be emitted as 'do not open'. Unbuyable tonight regardless: Quality_Pass False, MoS -7.29 -> val_adj -10. |
Outcome & track record
Accuracy at the time of this record.
28 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.