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 | 185.60 | SELL 30% | 185.60 | 169.53 | 185.61 (+0.0%) | 187.10 (+0.8%) | 187.31 (+0.9%) | KLA is a high-quality process-control franchise that just guided the September quarter to $4bn and lifted its 2026 wafer-equipment view, and the shares rallied 7.3% on the session — but the risk line under this position has walked away from it without the position ever being reduced, and that is a risk failure regardless of how good the business looks. We are banking a disciplined 30% and keeping in the name. ADR-0020 Trail Erosion is the sole surviving origination: live stop 160.55 vs a 272.00 Peak Stop set 2026-07-01 = 40.97% erosion with ZERO size reduction ever recorded; first fire on this watermark, veto-immune, pinned at the configured 30% floor tier. US-34/US-18 DECLINE the signal leg: today STRONG SELL - REDUCE cv76 Portion 88.52% (Sell_Rank 1) is a LONE settled flip out of AVOID cv63 x2 — no 2nd consecutive print, close 185.60 far above the 169.53 invalidation, and the name closed +7.32%. The 88.52% liquidation is explicitly declined. Valuation amplify (MoS -3.93 <= -3, would bump is NOT applied: the amplify rule scales a SIGNAL-originated trim, and the signal leg was vetoed — a risk-discipline mandate keeps its configured tier. This is the 5th consecutive session carrying a KLAC trim card unexecuted (09-01, 09-02, 09-03 x2 sizings, 09-04) — US-37 STALEMATE, needs an explicit Disposition (trade sell us KLAC 300 / trade skip us KLAC), not a silent 6th re-fire. Stop raised 160.55 -> 169.53 on the core. Forward view is dead (12wk 187.31 = +0.92% vs 185.60, T12_Progress 99.09%) which is why the core is trailed tight, not added to. · news |
| GOOGL | Alphabet, Inc.- Class A | 338.46 | SELL 30% | 338.46 | 335.34 | 338.30 (-0.0%) | 341.38 (+0.9%) | 382.16 (+12.9%) | Alphabet is up 175% on our cost and the engine has now gone silent on it for three straight sessions while the price stalls 0.9% above its own trailing stop. The news is genuinely good — the court rejected the DOJ AdX divestiture, Berkshire has made it a top-three holding, Cloud backlog is $460bn — so we are banking 30% of a triple and keeping long, not exiting the thesis. ADR-0012 armed + fired: PROFIT_ARMED with the arm FULLY US-26-VERIFIED (Avg_Cost 123.00, Unreal_PnL +175.17%, Gain_ATR 30.78 — no artifact), fired on tick (d) three consecutive NO TRADE cv0 prints (09-02, 09-03, 09-04). One tick -> profit_take_1 = 30%. Ticks (b) and (c) did NOT fire (12wk 382.16 = +12.9%, still alive) so this stays a 30% trim and never a full exit. US-17 is satisfied via the armed+fired path, not via the bare NO TRADE label. Valuation is inert here: MoS -0.63 > -3, no tier bump either way. News classified NEUTRAL — constructive but not a fresh dated catalyst, and it does not override a partial profit-take that leaves 70% of the position intact. 2nd consecutive session for this card (withdrawn 09-02, re-fired 09-03) — US-37 stalemate territory, needs an explicit Disposition. Core stop held at 335.34 (chandelier 2.0xATR = 324.46 is lower; ratchet forbids lowering). · news |
| NBIS | Nebius Group, N.V. | 226.39 | SELL 30% | 226.39 | 205.07 | 229.56 (+1.4%) | 261.71 (+15.6%) | 329.82 (+45.7%) | Nebius is compounding revenue at 454% year-on-year and closed +7.5% today, but it has funded that growth with 75bn of convertible senior notes and has an NVIDIA warrant over roughly 21m shares becoming exercisable on 09-11 — a leverage-plus-dilution stack sitting four sessions away, against a position whose stop has drifted 16% below its own high-water mark and has never been cut. Taking 30% off and keeping of a name whose 12-week view is still +45.7%. ADR-0020 Trail Erosion is the sole origination: live stop 197.16 vs 235.69 Peak Stop (2026-08-17) = 16.35% erosion, zero size reduction, first fire on this watermark, veto-immune. No signal origination exists — today prints AVOID cv57 with Portion 0. Valuation amplify (MoS -8.16) NOT applied for the same reason as KLAC: no signal-originated trim to scale. IMPORTANT: the fire stays OUTSTANDING until the position is actually cut. Stop is being re-based up to 205.07 (close - 1.5xATR) on the strength, which mechanically drops next session erosion to 12.99% — that re-base must NOT be read as the mandate evaporating; ADR-0020 immunity attaches to the un-cut state. 4th consecutive unexecuted session (09-01, 09-02, 09-03, 09-04) — US-37 STALEMATE, needs an explicit Disposition. Catalyst-aware note: 09-11 warrant is inside 5 sessions, which argues for taking the mandated tier NOW rather than pre-trimming larger into it. · news |
| TSM | Taiwan Semiconductor Manufacturing Co | 428.91 | BUY | 428.91 | 419.45 | 433.90 (+1.2%) | 441.75 (+3.0%) | 467.34 (+9.0%) | TSMC is the highest-conviction fresh buy on the board and the only STRONG BUY in the book tonight: revenue growing 41% with gross margin at 66%, a $265bn Arizona build and a first European fab funded, and capacity for leading-edge AI silicon still sold out. The stock reclaimed a failed breakdown today and closed at the day high with all three pillars aligned bullish and price 13.8% above its 200-day line. Buying a half-size starter at the open. Buy_Rank 1, conv 86, F-Score 9, Quality_Pass TRUE. Adjusted 86 -10 (valuation) -4 (FailedBreakdown_Up_20D|buy cell, n=11, hit_1w.30) = 72, comfortably clear of the 65 floor and the highest adjusted score in the pool. Sized at 0.5x on MoS -13.82: NOTE the EPV-based fair value assigns TSM an FV of against a price, and does the same to ASML (vs) and SNDK (vs) — this is a structural EPV artifact on high-growth semis, not a TSM-specific warning (systemic finding S10). The -10 val_adj is applied as the model computes it and is NOT hand-overridden. Not a chase: Entry_Distance_ATR -0.0, the model prints a Market order at its own entry. But the close IS the day high (428.91 vs 429.27) — REPRICE to the live entry if the Tuesday open gaps more than ~1.5%. Entry range 426.50-431.32 (+/-0.25xATR). 12wk 467.34 EXCEEDS the DCF anchor of by a wide margin — projection assumes growth beyond what is priced in. Stop 419.45 = -2.21%, a tight 1.0xATR risk line; Target_Price 448.19 is the trade objective, distinct from the 4wk estimate. · news |
| SAP | SAP SE | 215.11 | BUY — QUEUED @214.59 | 214.59 | 207.35 | 216.81 (+1.0%) | 224.75 (+4.7%) | 224.88 (+4.8%) | SAP has been reset hard — down 29.6% year to date and 40.5% over twelve months on a cut 2026 profit outlook — and is now rebuilding above its 200-day line with cloud revenue still guided to grow 23-25% and an AI Agent Hub launched. This is a bottoming trade with a defined risk line, not a momentum chase: buy the pullback into VWAP at 214.59, risk 3.4% to 207.35. The most stable buy in the pool: four consecutive BUY prints (71/64/76/71) on PullbackBuy_VWAP, the L1-favoured family and the only actionable buy cell with a positive bias (+1, n=25, hit_1w.542). Adjusted 71 -6 (valuation) +1 (cell) = 66, clears the 65 floor. Quality_Pass TRUE, F-Score 8. Entry_Distance_ATR -0.08, not extended; the 214.59 limit sits right on todays 214.47 low so it is genuinely fillable. Entry range 213.03-216.15. Sized 0.75x on MoS -1.27. NEWS: the 09-01 SAP card was demoted on news. Re-checked tonight and classified NEUTRAL, not a re-demote — the bearish material (downgrades, slower AI rollout, the profit-outlook cut) is the SETUP for a -40% reset, not a fresh dated catalyst on 09-03/09-04. Per the provenance discipline an override needs a dated catalyst; there is none. Flagging the reversal of the prior demote explicitly rather than silently. 12wk 224.88 exceeds the DCF anchor of — projection assumes growth beyond priced-in. GTC limit expires 2026-09-10 (anchor + 3 sessions; 09-07 is Labor Day). · news |
| LLY | Eli Lilly & Co | 1149.36 | HOLD | 1149.36 | 1168.46⚠ | 1148.48 (-0.1%) | 1167.30 (+1.6%) | 1140.23 (-0.8%) | Lilly is the only holding below its stop and the only one whose 12-week view has gone negative, but it is being HELD one more session on a line the book pre-committed to, not on hope. The GLP-1 franchise is intact; the recent softness is the 875bn Merida Biosciences deal plus continued Novo competition, neither of which is a thesis break. The pre-committed 09-03 trigger was a close below 1142.00 and it did NOT fire: LLY traded down to 1138.31 intraday but SETTLED at 1149.36, above the line. L4/US-22 — the settled close is authoritative — so the trigger is not fired and the pre-commitment is honoured rather than re-litigated. Both L2 legs ARE now satisfied (4th consecutive session below the 1168.46 stop, 12wk 1140.23 = -0.79% vs close) and LLY would otherwise be a dispose candidate; it ranks 4th and the three DISPOSE slots are taken by two mandatory ADR-0020 fires and one fired ADR-0012 profit-take. Both reasons are stated so this is not a silent pass. US-26 DECLINES the profit-take path: PROFIT_ARMED here is an ARTIFACT — no Avg_Cost, no Gain_ATR, no Unreal_PnL recorded, and the arm comes from Target_12Week (1140.23) collapsing BELOW the close, which is precisely the artifact US-26 names. Ticks (c) and (d) are therefore not usable as an origination. Stop of record deliberately LEFT at 1168.46 so the monitor keeps flagging the breach; the operative decisive line is 1142.00. A settled close below 1142.00, or a 5th session below stop with the 12wk still negative, fires a 30% trim automatically at the next EOD. Overlay override intact (invalidated only below 975). · news |
| NVDA | NVIDIA Corporation | 230.36 | HOLD | 230.36 | 220.48 | 231.81 (+0.6%) | 233.29 (+1.3%) | 238.44 (+3.5%) | NVIDIA is the strongest holding on the board — conviction 93 for a second straight session, all pillars aligned, price 15% above its 200-day — and it is being held in full with the trail left at 220.48. ADR-0012 would otherwise FIRE a 30% profit-take: PROFIT_ARMED plus tick (a) RSI_DIV, on an AnticipationUp_VCP signal so the G1 breakout protection does not apply. It is BLOCKED for a FOURTH consecutive session by US-26: no Avg_Cost is recorded for NVDA in input/usportfolio.json, so Gain_ATR and Unreal_PnL_Pct are both blank and the arm cannot be verified as required. The T12_Progress limb reads 96.61 with the 12wk target ABOVE the close, which is the non-artifact shape — but US-26 demands the gain legs, not judgement. FIX THE INPUT: adding NVDAs to input/usportfolio.json is worth more than any signal tweak this week — it is the single blocker on a live 30-share trim of a +RSI-divergent winner. Logged as systemic finding S7 for a fourth session. Trail held at 220.48; the models own chandelier (215.95) is lower and the ratchet forbids lowering. |
| MU | Micron Technology, Inc. | 1016.59 | HOLD | 1016.59 | 929.07 | 1030.05 (+1.3%) | 1167.13 (+14.8%) | 1371.62 (+34.9%) | Micron broke a thousand dollars a share today, closing +6.10% at 1016.59 and turning PROFIT_ARMED for the first time, with the 12-week view still projecting +34.9%. Holding all — this is the books biggest winner and there is no deterioration to act on. ADR-0012 ARMED (Gain_ATR 3.47 and Unreal_PnL +18.28%, both positive — arm fully US-26-verified) but NOT FIRED: no second Exit_Warning code, conviction flat, 12wk 1371.62 alive at +34.9%, no repeated NO TRADE. Armed-but-not-fired mandates a trail-raise, not a trim. TRAIL-RAISE DEFERRED, stated explicitly: the scans own PROFIT_ARMED-tightened chandelier prints 926.05 (2.0xATR on a 45.27 ATR = 4.5% of price), which is BELOW the standing 929.07, so the ratchet holds the existing stop and no numeric raise is available from the model tonight. The standing stop already satisfies the mandate. Overlay override intact: invalidated only if HBM pricing/demand cracks or on a decisive stop break. |
| OKTA | Okta Inc. | 170.60 | HOLD | 170.60 | 163.59 | 172.03 (+0.8%) | 182.35 (+6.9%) | 191.95 (+12.5%) | Okta is +22% on cost, trending well above every moving average, and turned PROFIT_ARMED today with the 12-week view still at +12.5%. Holding all on a raised-as-far-as-the-model-allows trail. ADR-0012 ARMED (Gain_ATR 3.22, Unreal_PnL +22.03% — arm US-26-verified) but NOT FIRED: conviction slipped only 2 points, far short of the 20-point collapse tick), no second Exit_Warning code, 12wk alive, and only ONE isolated NO TRADE print (09-02) rather than the two consecutive that tick (d) requires. Trail held at 163.59; the models chandelier prints 151.50, below the standing stop, so the ratchet applies. Note OKTA fails the fair-value quality gate (Quality_Pass FALSE) — that blocks NEW entries only, never a held name and never a sell. |
| NET | Cloudflare, Inc. | 278.92 | HOLD | 278.92 | 283.72 | 280.68 (+0.6%) | 282.58 (+1.3%) | 291.85 (+4.6%) | Cloudflare gave back 1.96% today and closed 278.92, back BELOW the 283.72 stop it reclaimed on 09-03 — the 08-21 self-heal has un-healed. Holding: the exit basis is not met and the 12-week view is still positive. L2 requires BOTH legs and only one is present — stop-break YES, but 12wk 291.85 = +4.6% vs close, so the negative-12wk leg fails. No trim. CRITICAL DEFECT (S9, now materially dangerous): input/usportfolio.json contains TWO NET rows — one Cost 274.25 and a later duplicate — and the scan resolves the duplicate, so the anchor CSV prints NET as Held=N. Consequence: the portfolio monitor did NOT report this stop breach; it surfaced only as an unledgered_drift warning. A held name breaching its stop invisibly is a monitoring failure, not a cosmetic one. DELETE the duplicate zero-NET row. |
| SPCX | Space Exploration Tech Corp | — | HOLD | — | 129.00 | n/a | n/a | n/a | SpaceX remains muted on signals — absent from the scan CSV entirely, no close, no action synthesised — but it is news-assessed as required and the read is CONFIRMS-HOLD. Held on the manual 129.00 stop against a 65.28 cost. News: SPCX IPO'd on NASDAQ 2026-06-11 at closed its debut +19% at raised $75bn against a 8tn valuation and now carries a ~$2tn market cap; consensus is Moderate Buy with an average 12-month target around -232 and a recent Oppenheimer target raise on subscriber growth. Nothing here argues for cutting a position at 65.28. ESCALATED SYSTEMIC FINDING (S8, upgraded): SPCX has been PUBLICLY QUOTED ON NASDAQ FOR NEARLY THREE MONTHS and the scan still produces no row for it. The permanent-mute rationale (private, no public quote) is now FACTUALLY STALE. This is a live 300-share position running with no signal, no conviction, no exit warning and no auto-unmute path. Add SPCX to the fetch universe and let the auto-unmute rule do its job. · news |
Outcome & track record
Accuracy at the time of this record.
185 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.