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 |
|---|---|---|---|---|---|---|---|---|---|
| CRM | SalesForce Inc | 243.00 | SELL 30% | 243.00 | 232.70 | 249.31 (+2.6%) | 264.04 (+8.7%) | 346.27 (+42.5%) | Salesforce broke its risk line decisively and has now closed below it for a third straight session, so the position gets cut by 30% while the long-term case stays intact. The business itself is not the problem — Agentforce ARR has crossed 5bn (+240% y/y), the combined Agentforce + Data 360 run-rate is about 9bn, and the 16 September Dreamforce Investor Day with the ClaudeForce launch is a genuine near-term catalyst — and at a DCF anchor the shares still screen as fair value rather than expensive. But a broken stop with conviction sliding is a risk failure regardless of the story, so we shrink the risk and hold 70% into the catalyst. || OPERATIONAL: this is a RE-FIRE, not a new decision. The identical 30% trim was decided on 2026-09-10 and never executed — the auto-executor's last run was 2026-09-09, so the market-on-open sell lapsed and the book still holds all. L3/UT-3/UT-11 mandate escalating an unexecuted exit to market-on-open rather than letting it lapse or re-recommending it indefinitely. Action price REPRICED from the stale 244.16 to the live anchor Entry_Price 243.00 per the provenance rule. || Origination (CLAUDE.md §9 signal-based): the 09-09 armed trigger fired on both legs (2nd consecutive close below the 251.77 stop AND a close below 245.99) and has since extended to a third consecutive close — 249.12, 244.16, 243.00. || TRIM not exit: L2's second leg is absent — 12wk 346.27 is +42.50% vs the action price and the DCF anchor 393.56 sits ABOVE that projection, so no growth-beyond-priced-in stretch is assumed. || No valuation amplify: live MoS_FV -1.88 is not <= -3, so no tier bump. || COUNTER-EVIDENCE, stated explicitly: the FailedBreakdown_Up_20D|sell cell is now ACTIONABLE at n=5, hit_1w 0.000, bias -10 — all five prior sells in this family were wrong at 1 week. That argument caps the trim at the 30% minimum tier; it does not cancel an unexecuted risk cut, because 'never got cut at all' is the failure ADR-0020 exists to prevent. || Residual, stop held at 232.70 (0.98 ATR under the close); the scan's 237.74 is a 0.50-ATR noise line and is rejected under this book's >=1.0-ATR stop-adoption convention. || Proceeds; realizes -. f=0.30 on the 1wk per L5 (calib_1w -0.163). · news |
| NET | Cloudflare, Inc. | 311.17 | NO ACTION | 311.17 | 302.52 | 321.55 (+3.3%) | 345.78 (+11.1%) | 368.02 (+18.3%) | Cloudflare is the best-ranked buy on the board and we are still not buying it, because the business does not yet fund itself. At Goldman's Communacopia conference the CFO lifted the long-run growth 'north star' from 40% to 50% a year and reframed the company as an agentic-AI platform rather than a software subscription — revenue is compounding at 36% — but it converts only of that into free cash flow against a market value north of $100bn, a 0.31% cash yield, on an F-Score of 4 and roughly ten times our valuation anchor. We will own it when the cash follows the narrative. || Mechanics: adjusted conviction 68 (raw 78, val_adj -10, bias 0) CLEARS the 65 floor and Entry_Distance_ATR is -0.0 so the no-chase bar is clear too — nothing in the technical read declines this. It is blocked solely by the hard quality gate (Quality_Pass False: F-Score 4, FCF, DCF anchor 29.51 vs a 311.17 close). Third consecutive session blocked on the same screen. || Ledgered as a counterfactual so the gate's cost gets priced (learnings advisory (d)(iii)). Re-arm: a Quality_Pass flip on refreshed fundamentals, or a genuine FCF inflection. · news |
| AMD | Advanced Micro Devices, Inc | 503.60 | NO ACTION | 503.60 | 493.33 | 515.92 (+2.4%) | 544.68 (+8.2%) | 507.88 (+0.8%) | AMD has the strongest fundamental case of any name we could buy today and it still falls seven points short of our entry bar. MI450 production shipments begin at the end of this quarter with what management calls a 'very significant step up' in Q4 and a further ramp through 2027, data-centre revenue is guided to double to $70bn in 2027, Anthropic has committed to up to 2 gigawatts of MI450 in Helios racks, and last quarter's data-centre revenue was a record 7bn — and unlike Cloudflare, Lumentum or Snowflake, AMD actually passes our cash-quality screen (F-Score 8, 4bn of free cash flow). What stops us is price: the shares sit about twelve times our valuation anchor, and that haircut takes the signal below the threshold at which we are willing to put money to work. || Mechanics: the label upgraded BUY - ANTICIPATE -> BUY (Buy_Rank 2) but conviction fell in the session's broad AI-hardware risk-off (AMD -3.36%), so adjusted conviction is 58 (raw 68, val_adj -10 from MoS_FV -11.22, bias 0) — 7 under the 65 floor. Not a chase (Entry_Distance_ATR -0.0). || Re-arm trigger: raw conviction >= 75 restores adjusted 65 and makes this an immediate ACQUIRE at the then-printed Entry_Price. || Counter-evidence to weigh at that point: this book's anticipate counterfactual is n=10, median 4wk -8.81% ['saved']. · news |
| LITE | Lumentum Holdings | 935.70 | BUY-STOP | 935.70 | 904.29 | 973.39 (+4.0%) | 1061.33 (+13.4%) | 1135.80 (+21.4%) | Lumentum's optical-networking transformation is real — FY26 revenue of 01bn is up 83% and it has won multi-year optical-circuit-switching business — but it turns almost none of that into cash ( of free cash flow, a 0.35% yield, F-Score 5) at roughly nine times our valuation anchor, so it stays off the book. Today also took the shine off the chart: the shares fell 5.4% and the signal dropped back to a watch-only reading. || Mechanics: the anticipate slot's count test is MET (BUY - ANTICIPATE in 2 of the last 3 snapshots: 09-09, 09-10) and conviction 61 clears the 55 slot floor — but the current row no longer prints an anticipation trigger (WATCH, conviction and Quality_Pass is False, so the slot is declined on both the hard gate and a dead trigger. Adjusted conviction 51 is 14 under the floor besides. || Ledgered as a counterfactual so the gate's cost gets priced. Re-arm: a fresh BUY - ANTICIPATE print plus a Quality_Pass flip. · news |
| DELL | Dell Technologies Inc | 506.62 | HOLD | 506.62 | 469.94 | 525.78 (+3.8%) | 570.47 (+12.6%) | 721.93 (+42.5%) | Dell gave back 5.4% in a broad AI-hardware pullback and remains the book's one large winner, up 9.8%, so we hold it with the stop unchanged rather than react to a single red day. The recurring worry on Dell is margin, not demand: rising memory prices land on the buyer of memory, and gross margin has already compressed from 21% to 18% as the mix shifts to lower-margin AI servers. Against that, the forward view is still +42% and nothing in the signal has broken. || Mechanics: conviction is a 16-point slide, short of the 20-point collapse that would be a profit-take fire tick; Gain_ATR 1.42 is below the 3.0 arm and T12 progress is 70.18%, so the position is NOT armed. Stop HELD at 469.94 — a 2.0x ATR chandelier off the 506.62 close computes to 442.77, well below the standing line, and the ratchet never lowers; the scan's printed 490.66 is only 0.50 ATR under the close and is rejected under this book's >=1.0-ATR adoption convention. Cushion 1.15 ATR. ADR-0020 clear (live stop == peak stop). || Position is 36.25% of equity and carries 155% of the book's total unrealized gain — stated plainly per CLAUDE.md §9, never flagged as a diversification risk. || No dated 09-10 catalyst found on either permitted source; the memory-cost margin article cited is June-dated background, not today's news. · news |
| MU | Micron Technology, Inc. | 977.41 | HOLD | 977.41 | 969.00 | 1005.98 (+2.9%) | 1072.63 (+9.7%) | 1121.41 (+14.7%) | Micron's story is intact — its entire 2026 high-bandwidth-memory output is already contracted, capacity is set to double by year-end, operating margins are running above 80%, and fiscal Q4 on 30 September is guided to $50bn of revenue — so we hold. What changed is the risk, not the thesis: the shares fell 4.9% and the signal's conviction dropped sharply, leaving the position sitting almost exactly on its stop. We therefore pre-commit to the cut rather than debate it next session. || Mechanics: conviction is a 22-point collapse, over the 20-point threshold — but profit-take fire tick (b) requires the position to be ARMED first and MU is not (Exit_Warning blank, Gain_ATR -0.24, underwater -1.17%), so nothing fires today. Stop HELD at 969.00; a 2.0x ATR chandelier computes to 882.19, below the line, and the ratchet never lowers. Cushion is only 8.41 = 0.18 ATR, i.e. sub-noise. || ARMED EXIT TRIGGER: any close below 969.00 fires a 30% DISPOSE next session, no further confirmation. This is the CRM template applied before the breach rather than after it. || Overlay carries an MU override ('HBM pricing/demand cracks or decisive stop break') but it is 68 days stale with no numeric delta and is NOT applied. ADR-0020 clear. FQ4 09-30 is absent from the horizon file. · news |
| TSM | Taiwan Semiconductor Manufacturing C | 428.03 | HOLD | 428.03 | 419.45 | 434.29 (+1.5%) | 448.88 (+4.9%) | 433.06 (+1.2%) | TSMC is held but it is the position with the least room left to run: the model's own 12-week view is now 433.06 against a 428.03 close, barely 1.2% of upside, which is effectively a dead forward view. We hold because the stop has not broken and because no profit exists to take — the position is down 2.3% — but this is the first name that would be cut if anything else deteriorates. || Mechanics: PROFIT_ARMED prints again (T12 progress 98.84%) and is INERT per UT-9 — Gain_ATR -0.96, an arm-on-loss mislabel, not a winner banking gains. ADR-0012 fire tick (c) IS satisfied (Target_12Week 433.06 < 1.02 x close 436.59) but a tick cannot fire an unarmed position, so nothing triggers. Conviction. Stop HELD at 419.45; the scan's 407.18 is lower and the ratchet does not follow it down. Cushion 0.82 ATR. ADR-0020 clear. || Watch item rather than an action: a conviction print below ~55, or any close below 419.45, converts this into a trim candidate. |
| AAPL | Apple | 326.57 | HOLD | 326.57 | 314.62 | 331.73 (+1.6%) | 343.76 (+5.3%) | 337.00 (+3.2%) | Apple was the only thing that worked today, up 3.6% while thirty of the book's thirty-two names fell, and it has rescued itself from the edge of its own stop — the position is now effectively flat on cost with the strongest signal conviction in the book. We hold, and the armed exit we set yesterday simply moves out of reach rather than being cancelled. || Mechanics: the 09-10 armed trigger (any close below 314.62) did NOT fire — the close is 326.57, now 1.39 ATR clear of the line versus 0.09 ATR yesterday. Conviction is the book's largest single-session gain. The forward view has come back to life: Target_12Week 337.00 vs 1.02 x close 333.10, so ADR-0012 tick (c) no longer fires (it did yesterday). PROFIT_ARMED prints (T12 progress 96.91%) and remains INERT per UT-9 — Gain_ATR -0.05, still the wrong side of zero. Stop HELD at 314.62: a 2.0x ATR chandelier computes to 309.38 (exactly the scan's printed line), which is BELOW the standing stop, and the ratchet never lowers. Deliberately not re-tightened to a fresh 1.0-ATR line — that is what produced the 0.09-ATR risk line this book spent three sessions complaining about. ADR-0020 clear. || Armed trigger stands: any close below 314.62 fires a 30% DISPOSE next session. |
| NVDA | NVIDIA Corporation | 218.36 | HOLD | 218.36 | 215.95 | 222.43 (+1.9%) | 231.91 (+6.2%) | 235.95 (+8.1%) | Nvidia drifted another 2.4% lower with the rest of the AI complex and sits 0.4 ATR above its stop with the forward view still intact, so we hold without changing anything. The position is small (2.6% of equity) and down 3.7%; there is no profit to protect and no broken risk line to respond to. || Mechanics: PROFIT_ARMED prints (T12 progress 92.55%) and is INERT per UT-9 — Gain_ATR -1.25, clearly underwater. Conviction is a 5-point drift, not the 20-point collapse of fire tick (b). Forward view alive: Target_12Week 235.95 vs 1.02 x close 222.73, so tick (c) does not fire. Stop HELD at 215.95; the scan's 204.81 is lower and the ratchet does not follow it down. Cushion 2.41 = 0.36 ATR — tight, and flagged, but with only a 5-point conviction drift and a live 12wk it does not warrant an armed trigger this session. ADR-0020 clear. || Horizon: Nasdaq-100 quarterly changes announced 09-11, triple witching 09-18, index rebalance effective 09-21. |
| PLTR | Palantir Technologies Inc. | 165.86 | HOLD | 176.52 | 167.07⚠ | 172.28 (-2.4%) | 187.26 (+6.1%) | 211.44 (+19.8%) | Palantir has closed just below its stop — the first such close, and only a fraction of a day's normal range below the line — so we arm the cut rather than take it, because the 12-week view is still +27% and the business is growing 85% a year. The pressure is valuation, not execution: the shares carry a price-to-sales ratio near 59x and have been caught in a broad repricing of expensive software, with a fresh sell-side downgrade framing the next move as a valuation reset. One more close below the line and we reduce. || Mechanics: close 165.86 vs stop 167.07 = a 0.72% / 0.17 ATR breach — a graze, not decisive. L2's second leg is ABSENT (Target_12Week 211.44 = +27.5%), so this is not an exit. Conviction. Stop HELD at 167.07; a 2.0x ATR chandelier computes to 151.45, far below, and the ratchet never lowers. || ARMED EXIT TRIGGER (the CRM template, pre-specified in the 09-10 card): a SECOND consecutive close below 167.07, OR any close below 158.65 (1.0 ATR under today's close), fires a 30% DISPOSE next session, no further confirmation. || ADD DECLINED: the printed entry 176.52 sits 1.48 ATR ABOVE the close — a pronounced chase, not a pullback — and this book's PullbackBuy_VWAP|buy cell is n=2 at 0.000/0.000 (report-only). UT-10 also bars pyramiding into a held name that is already underwater. ADR-0020 clear. · news |
Outcome & track record
Accuracy at the time of this record.
35 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 →
-0.163 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.