Recommendations
2 to acquire · 3 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 |
|---|---|---|---|---|---|---|---|---|---|
| KLAC | KLA Corporation | 170.89 | SELL 30% | 170.89 | 160.55 | 172.68 (+1.0%) | 175.70 (+2.8%) | 175.97 (+3.0%) | Cut a third of KLA on risk discipline, not on a view. The position's protective stop has walked from 272 down to 160 since July -- a 41% retreat of the risk line -- while the holding was never once reduced, so the amount of money exposed to a break has quietly grown all summer. The stock has stabilised (today's print flipped from a sell to a neutral oversold-reversal read) and analysts remain constructive, but a risk line that keeps moving away from an untouched position is a process failure regardless of the thesis. Taking 30% off re-bases the line and leaves long the AI-capex story. OPS: ADR-0020 / US-36 TRAIL EROSION FIRES -- peak stop 272.00 (2026-07-01) vs live 157.23 = 42.2% erosion, zero size reduction since the peak. FIRST FIRE ON THIS PEAK STOP: THE HELD-NAME SELL VETO STACK (US-17, US-18/34, US-22, US-30) EXPRESSLY DOES NOT APPLY -- including today's US-30 stabilization print. US-32 DE-ESCALATION: the 08-28 card's / 88.52% signal basis DISSOLVED (STRONG SELL - REDUCE cv69 -> HOLD / ReversalUp_Oversold cv55), so the card re-sizes and is now originated solely by ADR-0020 at the mandatory 30% tier. US-26: PROFIT_ARMED is an ARTIFACT here (Gain_ATR -0.71, -2.77%) and must not originate a profit-take -- declined by rule. Valuation amplify SUPPRESSED by G2 (Est_12W 175.97 / DCF 61.80 = 2.85x, anchor unreliable). US-37 STALEMATE: 5th consecutive unexecuted KLAC card; the 08-28 @175.54 would have realised more than today's close. Residual on stop 160.55 (model 157.23 REFUSED as looser, US-29). Peak stop re-bases to the live stop on fill. · news |
| NBIS | Nebius Group, N.V. | 199.54 | SELL 30% | 199.54 | 185.70 | 206.53 (+3.5%) | 222.10 (+11.3%) | 272.26 (+36.4%) | Trim 30% of Nebius because the hard 213 line we committed to in writing broke decisively -- the stock closed 199.54, some 6.3% below it, after three straight sessions of decay. The business case has actually improved (the 75bn convertible was upsized and closed, clearing the overhang, and Goldman lifted its target to), so this is a discipline trim, not an exit: stay long a company still growing revenue over 450%. The one forward-looking reason to be lighter is dated -- roughly 21m NVIDIA warrant shares become sellable on 11 September, a real supply event inside the holding period. OPS: decisive break of the PRE-COMMITTED hard 213 stop is the origination (CLAUDE.md S9 'Contrast' clause; G1/G2/G3 do not apply -- this is a loss-prevention trim on a -5.7% position, not a profit-trim). L2 does NOT fire on its own (Est_12W +36.4% is positive, not negative). US-17 respected: the bare NO TRADE cv0 label is NOT the basis -- the broken line is. US-22 sector-vs-idiosyncratic: no thesis crack, so HALVE-AND-KEEP-CORE sizing, taken at the CONSERVATIVE 30% tier rather than 50% because this is a single mild -3.29% session, not the multi-name washout US-22 was written for. VALUATION AMPLIFY SUPPRESSED BY G2: MoS_FV -8.16 would bump, but DCF_per_share is BLANK and Rev_Growth is 8.55x (hyper-growth) -- anchor unreliable, no tier bump taken. US-31: the 'keep core' overlay override is 59 DAYS STALE and carries no news-override force; the live 08-27 news is what supports the core, not the overlay. AUTO-WITHDRAW (US-22 self-heal): if NBIS reclaims 213.00 on the next settled close, this card is withdrawn. Residual on a reset stop of 185.70 (close - 1.0x ATR 13.83). · news |
| GOOGL | Alphabet, Inc.- Class A | 335.02 | SELL 30% | 335.02 | 333.68 | 337.65 (+0.8%) | 338.13 (+0.9%) | 371.35 (+10.8%) | Bank a third of a position that has tripled. Alphabet is up 172% on cost, has now made three lower closes in a row, and today's engine print turned outright bearish for the first time in a week. The news explains it: an AI brain drain -- chief scientist Jeff Dean gone after 27 years, Gemini 3.5 Pro delayed -- plus a softer-than-expected Q2 search line has taken the stock down 6% in a month against a 3% rise for the S&P. Cloud at +82% and a $514bn backlog are why we keep rather than exiting. OPS: ADR-0012 PROFIT-TAKE, ARMED + FIRED. Armed genuinely (Gain_ATR 30.3, +172.37%, T12_Progress 90.22) -- US-26 verified, this is a real winner, not an artifact. Fire tick (d): 2 consecutive settled NO TRADE prints (08-28, 08-31). Ticks (a)/(b)/(c) did not fire (no 2nd warning code; conviction rose; Est_12W 371.35 > 1.02 x 335.02 = 341.72). US-33 satisfied -- the arm is corroborated by a non-(c) tick. ONE TICK -> profit_take_1 = 30%. DECLINED PRINT, STATED PER US-34: the engine printed STRONG SELL - REDUCE cv63, Sell_Rank 1, 88.52 -- a LONE flip out of NO TRADE x2, so the 89% signal exit is DECLINED pending a 2nd consecutive settled print or a close below the 333.68 trail. US-19: soft-tick fire executes at the FIRST tier, no day-by-day escalation. G1 n/a (no fresh breakout); G3 satisfied (signal bearish); valuation amplify moot (MoS -0.63) and G2-suppressed anyway (Est_12W / DCF 145.29 = 2.56x). US-37: 3rd consecutive unexecuted GOOGL card -- the 08-28 @346.59 would have realised more. Ex-dividend on 09-04, immaterial. Residual on trail 333.68. · news |
| NFLX | Netflix, Inc | 80.81 | BUY — QUEUED @78.46 | 78.46 | 76.07 | 80.30 (+2.3%) | 79.15 (+0.9%) | 85.71 (+9.2%) | Bid for Netflix on a pullback, don't chase it. This is the highest-ranked buy in the book and it sits on the one entry pattern that has actually worked here -- a dip back toward the volume-weighted average price inside a recovering trend. The stock has climbed more than 20% off its July low of 67.60 with Pershing Square having built a 3.15m-share stake, while profit is at a record even though the shares sit a third below last summer's high. The offset is honest: Q2 revenue narrowly missed and Q3 growth was guided to just 11.7%, which is exactly why we bid 2.9% below the close rather than paying up. OPS: Buy_Rank 1, STRONG BUY cv76. Fusion 76 - 8 (val_adj, MoS_FV -1.55 x 5) + 1 (PullbackBuy_VWAP|buy, n=24) = adj 69, clears the 65 floor. QUALITY GATE PASSES (Quality_Pass True, F-Score 7). L1-FAVOURED FAMILY -- the only actionable buy family in this book. NO CHASE: Entry_Distance_ATR -1.14 (close 80.81 sits 1.14 ATR / 3.0% above entry) -> QUEUE, do not buy at market. PROVENANCE: 78.46 is the anchor snapshot's own printed Buy Limit, entry range 77.94 - 78.98 (78.46 +- 0.25 x 2.063 ATR). Sizing: 0.8549 x x 20% =; size modulator x0.75 (-3 < MoS_FV -1.55 <= -1) =; / 78.46 = = max_position_pct 60% and neither binding. Target_Price 83.24 is the risk objective, distinct from the 4wk estimate. AUTO-WITHDRAW (US-8/US-9): family leaves PullbackBuy_VWAP; conviction < 65; Quality_Pass flips False; expires_on 2026-09-04 unfilled. If NFLX gaps BELOW 78.46 do not fill at the gap -- reprice at the next EOD. · news |
| AAPL | Apple | 325.13 | BUY-STOP — QUEUED @323.17 | 323.17 | 299.10 | 324.57 (+0.4%) | 318.29 (-1.5%) | 334.83 (+3.6%) | A small, self-limiting starter in Apple ahead of its 9 September product event, the first under new CEO John Ternus, who took over on 1 September. The pattern is a volatility contraction -- the stock coiling tighter before it moves -- and the order only fills if the shares actually push through the pivot, so we are paying for confirmation rather than hope. Apple carries the cleanest balance sheet in the book, but its own forward model is flat, which is why this is a half-size position rather than a conviction buy. OPS: ANTICIPATE SLOT (4th slot, separate from the <=3 ACQUIRE). ELIGIBLE: BUY - ANTICIPATE in 2 OF THE LAST 3 settled snapshots (08-28 ANTICIPATE cv83 / 08-31 WATCH / 09-01 ANTICIPATE cv83), conviction 83 >= the 55 slot floor. This is the US-35 pre-commit paying off -- the 08-28 card wrote the trigger down and the second print arrived. Fusion 83 - 10 (val_adj, MoS_FV -2.97 x 5 clamped) + 0 (AnticipationUp_VCP|buy n=2, REPORT-ONLY, not applied) = adj 73. QUALITY GATE PASSES -- F-Score 9, the best in the book. Sizing: 0.6076 x x 20% =; x0.75 (-3 < MoS_FV -2.97 <= -1) x0.5 (anticipate) =; / 323.17 = = (1.5% of the book). WEAK FORWARD, STATED: Est_4W 318.29 is -1.51% BELOW the action price and Est_12W is only +3.61% -- this is why it is half-size and stop-triggered. ADVISORY (US-28 style, NOT blended): counterfactuals.anticipate n=7 median -3.46% at 1wk / -9.86% at 4wk [saved] -- this book has historically been better off NOT taking anticipation entries; cited, not applied. ANOMALY: the model prints Buy Stop @323.17 which is 0.60% BELOW the 325.13 close, so it will fill at the open rather than on confirmed strength -- CAP THE FILL AT 325.18 (top of the 321.16-325.18 entry range) to preserve provenance. AUTO-WITHDRAW: family leaves the anticipation set; conviction < 55; expires_on 2026-09-04 unfilled. · news |
| NVDA | NVIDIA Corporation | 217.44 | HOLD | 217.44 | 217.07 | 218.69 (+0.6%) | 219.78 (+1.1%) | 226.49 (+4.2%) | Hold Nvidia and cancel the trim we had pending. The reason we were going to lighten -- deteriorating momentum -- did not show up: conviction is broadly steady, the 12-week view is still positive at +4.2%, and the stock barely moved on a soft session for the group. The catch is that the trailing stop at 217.07 now sits just 0.17% under the price, so a single weak day takes the whole position out automatically. That is the trail doing its job, not a reason to pre-empt it. OPS: US-32 DE-ESCALATION -- the 08-28 pending SELL 30% card is WITHDRAWN, its basis dissolved. ADR-0012 ARMED BUT NOT FIRED: PROFIT_ARMED via T12_Progress 96.0, but NO tick fires -- (a) no 2nd warning code; (b) conviction = -7, short of the -20 bar; (c) Est_12W 226.49 > 1.02 x 217.44 = 221.79, forward alive; (d) no NO TRADE print. US-33: T12-armed name needs a non-(c) tick and has none. US-26 CANNOT BE VERIFIED -- no Cost is recorded for NVDA in input/usportfolio.json, so Gain_ATR / Unreal_PnL_Pct are blank and the arm cannot be confirmed as a real gain; DECLINED TO ORIGINATE rather than guessing. Mandatory trail-raise already satisfied: 217.07 (state) > 201.53 (model 2.0x ATR chandelier), so the model stop is REFUSED as looser per US-29. ACTION NEEDED: fill the NVDA Cost field. Horizon: AVGO reports after the close 09-02 (ecosystem read-through). |
| LLY | Eli Lilly & Co | 1160.00 | HOLD | 1160.00 | 1168.46⚠ | 1165.06 (+0.4%) | 1188.73 (+2.5%) | 1186.80 (+2.3%) | Hold Lilly through a hair's-breadth stop breach. The shares closed 1160, which is 0.72% under our trailing line -- a marginal miss, not a break, on a name whose 12-week view is still modestly positive and which was the only large holding to finish the session higher. We hold, and we mark the line where the exit case genuinely opens rather than reacting to noise. OPS: MARGINAL BREACH, HOLD per the US-22 MU precedent (-0.21% / -0.71% breaches held). L2 does NOT fire -- the combo requires a NEGATIVE 12wk and Est_12W 1186.80 is +2.31%. PROFIT_ARMED present (T12_Progress 97.74) but US-26 CANNOT BE VERIFIED -- no Cost recorded in input/usportfolio.json, so the arm cannot be confirmed as a real gain; DECLINED TO ORIGINATE. US-33 also bites: armed via the T12_Progress limb, and the only candidate tick is (b) conviction, which is a NoTrade cv0 artifact (US-29) rather than a genuine collapse -- one NO TRADE print, so tick (d) needs a second. DECISIVE-BREAK LINE PRE-COMMITTED: a settled close below 1142.00 (stop - 0.75x ATR 35.34) opens the exit case. Stop 1168.46 unchanged. ACTION NEEDED: fill the LLY Cost field. |
| MU | Micron Technology, Inc. | 933.44 | HOLD | 933.44 | 929.07 | 940.00 (+0.7%) | 1068.52 (+14.5%) | 1234.98 (+32.3%) | Hold Micron. It is up 8.6% on cost and carries the most attractive forward view in the book at +32% over twelve weeks, but the engine reads it as conflicted with no tradeable edge either way -- no case to sell, no case to add. The trailing stop at 929.07 is only 0.47% below the price, so the position is effectively self-managing from here. OPS: NO TRADE, NoTrade_Reason 'Consensus conflict + low conviction'. NOT profit-armed (Gain_ATR 1.66 < 3.0, T12_Progress 75.58 < 90) so the ADR-0012 path is unavailable. Trail 929.07 unchanged; the scan prints no stop on a NoTrade row. WATCH THE FORWARD: Est_12W has run 1101.42 -> 1210.28 -> 1234.98 across three sessions while the close went 932.86 -> 958.73 -> 933.44 -- a +12.1% projection drift on a flat tape (see systemic finding S6). Horizon: AVGO 09-02 after the close, TSMC August revenue 09-10. |
| OKTA | Okta Inc. | 166.43 | HOLD | 166.43 | 160.20 | 167.29 (+0.5%) | 176.87 (+6.3%) | 188.55 (+13.3%) | Hold Okta and nudge the stop up. The shares gave back 3.8% and round-tripped Monday's spike to overbought, but the position is still 19% ahead on cost with a +13% twelve-week view intact. Conviction sits below our action threshold, so there is nothing to add and nothing to cut -- just a slightly tighter line under it. OPS: HOLD cv56, below the 65 floor -- no add. Trail RAISED 160.05 -> 160.20 (model stop, which is higher, so it is accepted). NOT profit-armed (Gain_ATR 2.14 < 3.0, T12_Progress 88.27 < 90). Quality_Pass FALSE and MoS_FV -6.27, but valuation NEVER originates a sell (CLAUDE.md S9) and the gate blocks NEW ENTRIES ONLY, never a held name. Horizon: Goldman Communacopia 09-08, Adobe results 09-10 (group read-through). |
| SPCX | Space Exploration Tech Corp | — | HOLD | — | 129.00 | n/a | n/a | n/a | Hold SpaceX. It is privately held with no public quote, so it carries no signal and no price targets -- it is held on a manual stop and reviewed on news alone. The news is supportive: the latest insider share sale marked the company at roughly $800bn, close to double the $400bn mark set in July, and management is openly preparing for a possible 2026 listing on the back of Starlink and Starship. The one thing to diarise is a 7% employee-equity unlock on 10 September, which adds supply. OPS: permanent no_close mute -- absent from the scan CSV by construction, never synthesized; news-assessed every run per the eval-eod unpriceable-holding rule. Manual stop 129.00 unchanged. Auto-unmutes to full signal + fusion treatment the moment it appears in the scan with a non-null Close. Horizon: 09-10 employee-equity unlock (90-day tranche). · news |
| VIST | Vista Energy, S.A.B. de | 74.71 | NO ACTION | 73.81 | 65.00 | 75.04 (+1.7%) | 76.10 (+3.1%) | 80.37 (+8.9%) | Vista Energy is the single best-scoring name in the book and we cannot buy it yet. Conviction 86 is the highest of any candidate, it is the ONLY holding or candidate the valuation model calls outright undervalued, and it passes the quality screen -- but our anticipation slot requires the setup to print twice in three sessions and this is only its first. The trigger is written down here so the second print gets acted on tomorrow instead of rediscovered. OPS: ANTICIPATE SLOT INELIGIBLE -- 1 of the last 3 snapshots (NO TRADE / BUY / BUY - ANTICIPATE cv86); requires >=. Slot is in any case taken by AAPL this session. US-35 PRE-COMMITTED TRIGGER: if VIST prints BUY - ANTICIPATE again on the 09-02 settled close and conviction stays >= 55, QUEUE a GTC BUY-STOP at the then-printed Entry_Price, sized 0.6076 x x 20% x 1.25 (UNDERVALUED + Quality_Pass) x 0.5 (anticipate slot) = This is the SECOND consecutive session VIST has been slot-blocked (also 08-01) -- flagged as systemic finding S5. Val_Signal UNDERVALUED, MoS_FV +0.0014, F-Score 7, DCF 86.44 vs a 74.71 close. |
Outcome & track record
Accuracy at the time of this record.
171 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 →
-0.191 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.
52 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.