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 | 182.91 | SELL 30% | 182.91 | 174.01 | 183.94 (+0.6%) | 194.41 (+6.3%) | 183.73 (+0.4%) | KLA is a profitable position whose protective stop has walked down 36% from its peak while we have never once reduced the size - that is a risk failure, not a view on the company, and we are cutting 30% to fix it. The forward view has also gone flat (12-week estimate sits +0.5% from here) and the stock is priced well beyond any defensible intrinsic anchor, so there is little being given up. The remaining stay long. OPERATIONAL: ADR-0020 Trail Erosion origination - live trail 174.01 vs a 272.00 Peak Stop (2026-07-01) = 36.03% erosion, zero size reduction since that peak. FIRST fire on this Peak Stop, so it is VETO-IMMUNE (US-36): US-17/US-18/US-34/US-22/US-30 do not apply, and neither do the G1/G2/G3 trim-discipline gates nor the catalyst-aware hold (those govern the valuation-elevate path only). Signal leg is AVOID cv56 and is NOT the basis - the 09-04 STRONG SELL-REDUCE cv76 lapsed as a US-34 lone flip and never repeated. Valuation amplify WITHHELD despite MoS -4.28 <= -3: CLAUDE.md section 9 amplify attaches only to signal-originated sells, and there is none tonight - pinned at the ADR-0020 30% floor. SECOND consecutive session carded unexecuted = US-37 STALEMATE; re-carded at the SAME tier, no US-19 escalation. Model Stop_Price 198.55 prints ABOVE the 182.91 close - US-29 bearish-family artifact, ignored. Core trail HELD at 174.01 (2.0xATR chandelier computes 167.27, lower - ratchet holds). Peak Stop re-bases to 174.01 on execution and any future fire loses veto-immunity. Proceeds. · news |
| LLY | Eli Lilly & Co | 1124.21 | SELL 30% | 1124.21 | 1066.29 | 1125.07 (+0.1%) | 1122.92 (-0.1%) | 1114.05 (-0.9%) | Lilly broke the line we committed to in advance and has not reclaimed it in six sessions, and every forward estimate we have - one week, four week, twelve week - now sits flat to slightly below today's price. We are taking 30% off and keeping the rest; this is a stalled position being sized down, not a thesis being abandoned. OPERATIONAL: origination is US-27 / L2 - the decisive break of the PRE-COMMITTED 1142.00 line (written 09-03) remains unreclaimed at a settled 1124.21, sixth consecutive session under the original 1168.46 stop, and the 12wk estimate 1114.05 is -0.90% vs close, satisfying L2's negative-forward limb. The exit case was re-verified at this settled close per US-27 and did NOT auto-withdraw. US-26 DECLINES the ADR-0012 profit-take path: Avg_Cost is null in input/usportfolio.json so Gain_ATR and Unreal_PnL_Pct are unverifiable and PROFIT_ARMED here comes from the T12_Progress 100.91 artifact limb (Target_12Week collapsed BELOW the close). US-17 is not implicated - the NO TRADE cv0 label is not the basis either. ADR-0020 does NOT fire: trail 1061.36 vs a 1180.30 peak = 10.08% erosion, under the 15% threshold. Valuation gives no tier bump (MoS -2.58 is above the -3 line). SECOND session carded unexecuted = US-37 stalemate; re-carded at the SAME 30%. Core trail RAISED 1061.36 -> 1066.29 (2.0xATR chandelier). Proceeds. · news |
| GOOGL | Alphabet, Inc.- Class A | 330.65 | SELL 30% | 330.65 | 315.44 | 329.41 (-0.4%) | 335.40 (+1.4%) | 375.36 (+13.5%) | Alphabet closed below the stop we were protecting it with and the model flipped to its strongest sell reading, ranking it the number-one name to reduce. The antitrust relief that drove the stock up 9% a week ago is now fully in the price and the shares have given most of that move back. We are cutting 30% on the broken risk line while leaving to run, because the twelve-week view is still solidly positive. OPERATIONAL: dual basis - (1) stop breach, settled 330.65 vs the recorded 335.34 trail = -1.40%, the day's ONLY monitor breach; (2) active STRONG SELL - REDUCE cv63, Sell_Rank 1, Portion 88.52%. US-18/US-34's lone-flip bar (this is a single flip out of 3x NO TRADE cv0) is cleared via its INVALIDATION limb - a close below the trail - not via a 2nd print. The prior 09-08 card's basis (ADR-0012 armed + tick (d)) has DISSOLVED (cv, no longer NO TRADE), so per US-32 the card is not carried on a stale basis; it is RE-ORIGINATED on the harder stop-break + REDUCE basis at the SAME 30%. US-19: the model's 88.52% portion is DECLINED, no day-by-day escalation. Arm is US-26-genuine (Avg_Cost 123.00, +168.82%, Gain_ATR 27.30) but NO ADR-0012 tick fires today - (a) no 2nd warning code, (b) cv rose, (c) 12wk 375.36 > 1.02x close, (d) not NO TRADE - so this is a signal/stop sell, not a profit-take. ADR-0020 does not fire (335.34 vs 358.00 peak = 6.33%). Valuation gives no bump (MoS -0.61, the least-overvalued held name in the book). Scorecard RallySell_BearTrend|sell is -2 (n=13, hit_1w.385, hit_4w.222) - a weak sell cell, which argues for the minimum tier, and 30% is the minimum. Core trail re-based 335.34 -> 315.44 (model) with a PRE-COMMITTED hard line at EMA200 327.85: a 2nd consecutive settled close below 327.85 escalates the exit case at that EOD. Live event: 09-16 redaction-motion deadline on the sealed ad-tech opinion. Proceeds. · news |
| NET | Cloudflare, Inc. | 314.18 | ADD — QUEUED @309.52 | 309.52 | 292.77 | 314.66 (+1.7%) | 321.84 (+4.0%) | 350.54 (+13.3%) | Cloudflare is the strongest buy signal on the board and the news behind it is real - an OpenAI partnership, 36% revenue growth with raised guidance, and federal certification now used by over 100 agencies. But it jumped 10.5% in a single session and closed within 1% of its high, and this book's own record says chasing that kind of move is how it loses money. We are adding half the normal size to an existing position, and only on a modest pullback. OPERATIONAL: adjusted conviction 78 raw -10 (val, MoS -9.65) -3 (FailedBreakdown_Up_20D|buy cell, n=13 hit_1w.364) = 65, EXACTLY at the floor. Raw 78 clears the US-10/US-16 chase-family bar of 75. Quality gate does NOT block - Quality_Pass False (F=4) blocks NEW entries only and this is an ADD to a held at avg 274.25 (+14.6%); the scan reads Held=N only because of the S9 duplicate-row bug. QUEUED rather than market on L6/US-9: 09-09 gapped open 295.50 from a 284.30 close and finished 314.18, 1.05% under the 317.51 high - a news-gap session. Limit set at Entry_Price 314.18 minus 0.25xATR (18.62) = 309.52, inside the action-price provenance band. Sizing: Portion 49.24% x = 98, x0.5 size modulator (MoS -9.65 <= -3) = = Auto-withdraw if conviction drops under 65 or the family leaves FailedBreakdown_Up (US-8), else expires 2026-09-14. Held trail RAISED 283.72 -> 292.77. 12wk 350.54 exceeds the 29.51 DCF anchor - projection assumes growth beyond priced-in. Catalysts tonight: Oracle and Adobe results both read through to this name. · news |
| DELL | Dell Technologies Inc | 535.25 | BUY-STOP — QUEUED @541.74 | 541.74 | 443.77 | 541.50 (-0.0%) | 543.36 (+0.3%) | 669.76 (+23.6%) | Dell has signalled an imminent breakout for three straight sessions and is the only such candidate that passes our fundamental quality screen. The business case is exceptional - record quarterly revenue up 58%, AI server orders of 9bn, a $95bn backlog and full-year earnings guidance raised by 42% - and the stock has spent five sessions building a base rather than running away. We are placing a small starter order that only executes if the stock confirms by trading higher. OPERATIONAL: ANTICIPATE slot (4th slot, separate from the <=3 ACQUIRE). Eligible on 3-of-3 last snapshots (09-04 cv83 -> 09-08 cv63 -> 09-09 cv70); adjusted 70 -10 (MoS -2.48) +0 (AnticipationUp_VCP|buy n=4, report-only) = 60, clears the 55 floor - ASSUMPTION RESTATED: floor read on the ADJUSTED scale, same as 09-08. Quality gate PASSES (Quality_Pass True, F=7) - the only anticipate-eligible name that does; LITE (cv93) and ASX (cv88) both outrank it on conviction and are both BLOCKED by the gate. Sizing: 60.76% x x 20% = x0.75 size modulator (-3 < MoS -2.48 <= -1) = x0.5 anticipate = / 541.74 = = Buy-STOP fills only on confirmed strength above 541.74 - self-limiting; 09-09 printed a 562.99 high and closed 535.25, a failed push, so the trigger is live but unmet. Not an L6 gap-chase: the 09-02 post-earnings +16% print is five sessions and a base behind us. PRE-COMMITMENT (US-35): fills at 541.74, withdraws on conviction < 55 or a family change off AnticipationUp_VCP, else expires 2026-09-14. 12wk 669.76 exceeds the 446.05 DCF anchor. ADVISORY FLAG, cited not applied: counterfactuals.anticipate n=11 median wh_1w -3.46% [SAVED] - historically NOT queueing the anticipate has been better at 1wk on this book. · news |
| NVDA | NVIDIA Corporation | 223.67 | HOLD | 223.67 | 220.48 | 225.16 (+0.7%) | 224.83 (+0.5%) | 233.11 (+4.2%) | A profit-taking trim on Nvidia is triggered on the signals but we cannot act on it, because the position has no recorded purchase price in the holdings file and we refuse to bank a 'profit' we cannot verify exists. This is the sixth session running that a one-line data gap has blocked a valid decision. Holding on a tight stop. OPERATIONAL: ADR-0012 tick (d) FIRES - two consecutive NO TRADE cv0 prints (09-08, 09-09) - and T12_Progress reads 95.95. But US-26 BLOCKS the origination for a 6th session: Avg_Cost is null because the NVDA row in input/usportfolio.json carries no Cost field, so Gain_ATR and Unreal_PnL_Pct are both unverifiable and the arm cannot be confirmed genuine rather than the collapsed-target artifact. US-33 would additionally require a non-(c) tick, which (d) satisfies - so the ONLY thing standing between this and a mandatory 30% card is one missing JSON field. Systemic finding S7 escalated again. Trail HOLDS at 220.48 - chandelier 223.67 - 2x6.85 = 209.97 is lower, ratchet holds; close 223.67 sits 1.4% above the stop, the tightest held name in the book. ADR-0020 does not fire (peak = live = 220.48). |
| MU | Micron Technology, Inc. | 1027.77 | HOLD | 1027.77 | 953.77 | 1031.81 (+0.4%) | 1126.43 (+9.6%) | 1405.29 (+36.7%) | Micron is the strongest held name in the book - conviction rose again today and the twelve-week view points 37% higher. We are not taking any profit here; we are raising the protective stop by a share so more of the gain is locked in while the position keeps running. OPERATIONAL: ADR-0012 ARMED but NOT FIRED. Arm is US-26-genuine (Avg_Cost 859.49, Unreal +19.58%, Gain_ATR 3.66 - clears the 3.0 threshold on the real gain limb, not the T12 artifact limb, so US-33 is not implicated). Tick re-run: (a) no second warning code; (b) conviction ROSE, nowhere near a 20-pt collapse; (c) 12wk 1405.29 vs 1.02 x 1027.77 = 1048.33, forward alive at +36.7%; (d) not NO TRADE. Zero ticks -> mandatory trail-raise. Stop RAISED 929.07 -> 953.77 (model Stop_Price, which for once sits ABOVE the standing line - a genuine ratchet, not the US-29 loosening artifact). ADR-0020 silent (peak was 929.07, now re-based up). Signal is a Buy Stop @ 1045.66 (+1.74%) - not taken, the ACQUIRE slots went elsewhere and MoS -1.38 would haircut it to adj 76 with the position already at 14% of marked book. |
| NBIS | Nebius Group, N.V. | 240.35 | HOLD | 240.35 | 234.32 | 242.86 (+1.0%) | 262.48 (+9.2%) | 341.26 (+42.0%) | Nebius gave back a little of last week's Palantir-driven jump and its signal conviction dropped sharply, but the position is not near its stop and the twelve-week view is still 42% higher. Oracle reports after tonight's close and is the single biggest read-through for this name, so we hold through the event rather than pre-trimming into it. OPERATIONAL: HOLD cv61 (down 30 from 91). A 30-pt conviction drop is NOT an ADR-0012 tick because the position is NOT ARMED - no PROFIT_ARMED flag, Gain_ATR 1.84 under the 3.0 threshold, T12_Progress 70.43 under 90 - and tick (b) only exists on an armed name. No stop breach (240.35 vs 234.32, +2.6% of headroom). ADR-0020 silent: trail 234.32 vs a 235.69 peak = 0.58% erosion. Trail HOLDS at 234.32 - the model prints 232.53, lower, so the ratchet is not loosened (US-29). Catalyst-aware hold applies: Oracle FY27 Q1 after the close 09-10 (high confidence, the OCI backlog read-through), then CPI 09-11 and FOMC 09-16. Valuation is uninformative here - MoS -364.66 on a 0.66 FV_per_share is the S10 EPV collapse on a hyper-growth name, and the G2 DCF-reliability gate would suppress any amplify off it in any case. |
| SPCX | Space Exploration Tech Corp | 147.55 | HOLD | 147.55 | 140.06 | 146.83 (-0.5%) | 147.49 (-0.0%) | 162.60 (+10.2%) | SpaceX absorbed its 90-day lock-up expiry - the session where up to 319 million shares became eligible to sell - with a 3.9% dip and no break of our protective level. The position is still up 126% and the model reads it as a hold. We keep all and keep the stop where it is. OPERATIONAL: second session since the S8 auto-unmute; full signal treatment, no longer a news-only assessment. ADR-0012 ARMED (US-26-genuine: Avg_Cost 65.28, Unreal +126.02%, Gain_ATR 11.80 - the real-gain limb) but NOT FIRED: (a) no second warning code; (b) conviction ROSE; (c) 12wk 162.60 > 1.02 x 147.55 = 150.50, forward alive at +10.2%; (d) 09-08 was AVOID, not NO TRADE. Zero ticks -> trail-raise only, and the 2.0xATR chandelier computes 133.61, BELOW the standing 140.06, so the ratchet HOLDS at 140.06. ADR-0020 silent (peak = live). The Day-90 underwriter lock-up tranche was the medium-confidence 09-09 horizon event and it has now passed without a technical break. Valuation MoS -67.13 on a 2.17 FV_per_share is the S10 EPV collapse - meaningless for a pre-IPO-structure name, val_adj 0 by G2. |
| WDS | Woodside Energy Group Ltd | 24.19 | NEAR-MISS | 24.43 | 22.35 | 24.33 (-0.4%) | 24.88 (+1.8%) | 26.08 (+6.8%) | Woodside was the only Strong Buy on the board and the closest thing to a genuine miss tonight - a clean 52-week breakout on the one buy family whose track record here is actually positive. It fails our fundamental quality screen, so we pass. OPERATIONAL: adjusted conviction 78 -5 (MoS -0.98 x 5) +2 (BreakoutUp_52W|buy, n=11 hit_1w.60) = 75, comfortably clear of the 65 floor, and BreakoutUp_52W is NOT one of the three US-16 momentum-chase families so no raw-conviction bar applies. The BLOCK is the hard quality gate: Quality_Pass False, F-Score 4, no resolvable DCF. Buy Stop @ 24.43 (+0.99%), Entry_Distance_ATR +0.36 - not extended. Re-arms on a quality-screen pass at the next fairvalue refresh. |
| BAC | Bank of America | 62.67 | NEAR-MISS | 62.67 | 61.42 | 63.18 (+0.8%) | 63.93 (+2.0%) | 64.46 (+2.9%) | Bank of America clears our conviction bar and is barely above fair value by the standards we apply to banks, but it fails the fundamental quality screen and its own forward estimates only point 3% higher over twelve weeks. Not worth a slot. OPERATIONAL: adjusted 75 -2 (MoS -0.45) -3 (FailedBreakdown_Up_20D|buy) = 70, above the floor, and raw 75 exactly meets the US-10 chase-family bar. BLOCKED by the quality gate (Quality_Pass False, OVERVALUED_VS_RI on the residual-income bank method, F-Score 7). Buy_Rank 4. |
| TSLA | Tesla Inc | 367.81 | NEAR-MISS | 356.20 | 337.99 | 370.29 (+4.0%) | 362.66 (+1.8%) | 377.46 (+6.0%) | Tesla is on the one entry pattern this book trusts and would land exactly on the buy threshold, but it trades at roughly twenty times any defensible intrinsic value, fails the quality screen, and is still below its 200-day trend line. Pass. OPERATIONAL: adjusted 74 -10 (MoS -19.37, clamped) +1 (PullbackBuy_VWAP|buy, n=28 hit_1w.542) = 65, exactly at the floor - and BLOCKED anyway by the quality gate (Quality_Pass False, F-Score 4). Corroborating demotion: close 367.81 sits under EMA200 381.16, so the pullback is into a downtrend, not a trend continuation. Buy Limit @ 356.20, Entry_Distance_ATR -0.74. Buy_Rank 5. |
| LITE | Lumentum Holdings | 988.98 | NEAR-MISS | 1006.49 | 816.44 | 1003.86 (-0.3%) | 1075.02 (+6.8%) | 1401.44 (+39.2%) | Lumentum had the highest conviction reading of any anticipation candidate and a twelve-week view 39% higher, but it fails the fundamental quality screen, so the single anticipation slot went to Dell instead. OPERATIONAL: anticipate-eligible on 2-of-3 (09-08 cv100, 09-09 cv93); adjusted 93 -10 (MoS -7.75) = 83, far above the 55 slot floor and the top-ranked eligible name. BLOCKED by the quality gate (Quality_Pass False, F-Score 5) - the gate applies because this is a NEW entry. Buy Stop @ 1006.49. Together with ASX this is US-20 recurring: the quality gate x growth-priced regime keeps vetoing the highest-conviction entries the book generates. |
| ASX | ASE Technology Holding | 41.20 | NEAR-MISS | 40.59 | 35.81 | 41.47 (+2.2%) | 42.34 (+4.3%) | 48.60 (+19.7%) | ASE Technology is the second-highest conviction anticipation candidate with a twelve-week view 18% higher, and it too fails the fundamental quality screen. Passed over for the same reason as Lumentum. OPERATIONAL: anticipate-eligible on 2-of-3 (09-08 cv83, 09-09 cv88); adjusted 88 -10 (MoS -10.85, clamped) = 78, well above the 55 floor and ranked 2nd. BLOCKED by the quality gate (Quality_Pass False, F-Score 6, no resolvable DCF). Buy Stop @ 40.59 sits BELOW the 41.20 close (Entry_Distance_ATR -0.38), so it would trigger immediately - a further reason to want the gate to be right. |
| SNDK | Sandisk Corporation | 1764.17 | NEAR-MISS | 1764.17 | 1714.42 | 1759.48 (-0.3%) | 1641.54 (-7.0%) | 1760.53 (-0.2%) | SanDisk is the only buy candidate on the board that is actually near fair value AND passes the quality screen - a rare combination in this book right now - but its conviction is too low and its own four-week estimate points 7% lower. Not tonight. OPERATIONAL: adjusted 69 -7 (MoS -1.38) -3 (FailedBreakdown_Up_20D|buy) = 59, six points under the 65 floor; raw 69 also misses the US-10 chase-family bar of 75. Quality_Pass True (F-Score 8) and FAIR_VALUE - if the family were PullbackBuy_VWAP this would be the pick. Ledgered so the counterfactual grader prices the pass. Re-arms on conviction 76+ or a family flip to PullbackBuy. Buy_Rank 6. |
Outcome & track record
Accuracy at the time of this record.
201 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 →
-0.171 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.