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 | 208.25 | SELL 30% | 208.25 | 191.24 | 208.94 (+0.3%) | 215.37 (+3.4%) | 212.51 (+2.0%) | Trimming 30% of a very large winner purely on risk discipline, not on the business. KLA is doing well - it beat on Q2, raised its dividend, and rallied 3.9% today with the whole semi-cap complex on a cooler July CPI print - and the we keep stay long every bit of that. What we are fixing is that this 1000-share position has never once been reduced while its protective stop walked from 272.00 down to 182.81, a 33% erosion of the risk line with zero size taken off. That is ADR-0020 Trail Erosion and it is mandatory. The forward view also makes this the cheapest moment to do it: the model's own 12-week estimate is only 2% above today's price with 98% of the target already banked. ADR-0020 first fire, veto-immune under US-36, tier fixed at the 30% floor. Session 11 unexecuted, re-priced 200.47 -> 208.25 (a re-price, NOT a US-19 escalation). Erosion -32.79% vs Peak Stop 272.00 (2026-07-01, post-split); -29.69% even against tonight's raised 191.24 line. ADR-0012 profit-take tested and DECLINED: armed via T12_Progress 98.0 limb (Gain_ATR 2.87 < 3.0, US-26 passes on Avg_Cost 175.75 / +18.49%), but (a) no 2nd Exit_Warning code, (b) conviction is a label-flip RISE not a decay, (c) Target_12Week 212.51 vs 1.02xClose 212.415 = alive by and US-33 bars the (c)-only path on a T12-limb arm, (d) AVOID not a 2nd NO TRADE. Valuation amplify (MoS_FV -5.68 <= -3, would bump SUPPRESSED by G2: Target_12Week 212.51 vs DCF_per_share 83.43 = 2.55x overshoot. Catalyst-aware hold considered and inapplicable - AMAT FQ3 AMC 08-13 is a peer read-through, not KLAC's own event, and the clause governs the Exit_Warning-elevate path, not the ADR-0020 class. Provenance: action price = anchor settled close 208.25 (Entry_Price 208.25, EDist 0.0); printed Stop_Price 230.94 is a bearish-family invalidation ABOVE the close and is not used; the dead 07-01 @280 GTC is not republished. Stop on remainder 182.81 -> 191.24 (1.5 ATR). Keep; proceeds; realized ~+ (+18.49%). Scorecard bias -3 recorded for transparency (FailedBreakout_Down_20D|sell, n=20, hit_1w.368) but does NOT gate a risk-discipline card. · news |
| NBIS | Nebius Group, N.V. | 259.20 | SELL 30% | 259.20 | 215.37 | 259.63 (+0.2%) | 276.64 (+6.7%) | 296.04 (+14.2%) | Nebius just delivered a spectacular quarter - revenue up 454% year-on-year to, AI-cloud revenue up 514%, annualised recurring revenue up 56% in a single quarter to $3bn, and the year-end contracted-power target raised to 5GW - and the shares closed up 34%. We are keeping 210 of long every word of that. The coming off are not a verdict on the business; they are the overdue correction of a risk failure. This position has never been reduced by a single share while its stop sat 87 points below the peak we once protected - a 32% erosion of the risk line. We committed in writing to fix that at this exact session regardless of how the print landed, precisely so a good result could not talk us out of it. The stop also gets its largest raise of the run, from 183 to 215.37. ADR-0020 first fire, veto-immune (US-36), UNCONDITIONALLY pre-committed on 08-12 to the 20260813_06* anchor. Erosion -32.22% vs Peak Stop 270.00 (2026-06-23); still -20.23% against tonight's raised 215.37 line, so the fire is robust to the raise. Zero size reduction ever on. ADR-0012 NOT armed at all (Gain_ATR 1.63 < 3.0, T12_Progress 87.56 < 90, no Exit_Warning code) so the profit-take path is unavailable and cannot be the basis. Valuation amplify (MoS_FV -6.03 <= -3, would bump SUPPRESSED by G2: DCF_per_share is NA and Rev_Growth 8.546 (+855%) is the hyper-growth/unreliable-anchor case G2 exists for. Provenance: action price = anchor settled close 259.20; the row's Entry_Price 233.22 is a BUY-stop pivot and Stop_Price 145.57 a stale long stop - neither used for a sell. REPRICE WARNING: after-hours printed 253.50 (-2.20%); a >1.5% gap at the open reprices this card (L8). Stop on remainder 183.00 -> 215.37 (1.5 ATR, +17.7%); model 145.57 rejected as far looser (US-29). Keep; proceeds; realized ~+ (+22.50%). Est 12wk band [149.77-698.72] is +-4.7x wide - the honest statement of model uncertainty here; DCF annotation impossible (DCF NA). PRE-COMMITMENT (1) CARRIES, RE-BASED: any settled close below the new 215.37 stop, or a 2nd consecutive settled STRONG SELL - REDUCE, fires a mandatory 50% trim (105 of the remaining 210). · news |
| GOOGL | Alphabet, Inc.- Class A | 343.54 | SELL 30% | 343.54 | 327.40 | 345.32 (+0.5%) | 358.28 (+4.3%) | 384.28 (+11.9%) | Alphabet has broken the risk line we drew for it, twice. We committed on 08-08 that a settled close at or below would trigger a 30% trim, and the stock has now closed below it two sessions running (343.80, then 343.54) while the signal itself deteriorated from a neutral HOLD into an outright bearish AVOID. It is now trading below its 10-, 20- and 50-day averages and, tellingly, it closed red on a day when the wider market rallied on cool inflation data. This is a de-risk on the book's largest winner - the position is still up 179% and the 12-week view is still positive at +11.9%, so stay - not an exit from the Alphabet story. Carried from 08-12 unfilled, re-priced 343.80 -> 343.54. Decisive stop-loss break is a valid signal-based exit under CLAUDE.md section 9 and is EXPLICITLY EXEMPT from G1/G2/G3; US-34's bar met twice on the close-below-invalidation limb. Trail A07C NO TRADE -> A10C HOLD cv56 -> A11C HOLD cv52 -> A12C AVOID cv50, family flipped FailedBreakdown_Up_20D -> FailedBreakout_Down_20D. Close below EMA10 351.32 / EMA20 351.19 / EMA50 353.25, +6.2% over EMA200 323.60, RSI 45.73, VWAP_Distance -0.95. TIER HOLDS AT THE PRE-COMMITTED 30% - not escalated on the label degradation (US-19), not de-escalated (US-32). NOT the L2 combo: the 12wk is POSITIVE (384.28, +11.86%), so this is a de-risk, not a thesis exit. ADR-0012 armed genuinely (Avg_Cost 123.00, Gain_ATR 20.49, Unreal_PnL +179.30%, T12_Progress 89.40, PROFIT_ARMED) with ZERO ticks: (a) no 2nd code, (b) = -2, (c) 384.28 vs 1.02xClose 350.41 alive, (d) AVOID not NO TRADE - the profit-take path is not the basis. Valuation MoS_FV -0.74 too shallow to amplify and G2 would suppress (DCF 145.29 = 2.64x overshoot). Provenance: Entry_Price 343.54, EDist -0.0; printed Stop_Price 365.07 is a bearish-family invalidation above the close, not used. Stop on remainder 325.72 -> 327.40 (1.5 ATR), sitting +3.8% over EMA200 323.60. Keep; proceeds; realized ~+ (+179.3%). News neutral-to-mildly-constructive (Gemini 1B users, Pixel 11 launch, Brin AI push) with the 08-11 DeepMind exodus not compounded - no override. Per US-31 the 39-day-stale overlay override ('<345') carries no originating force. Scorecard bias -3 recorded; does not gate a pre-committed card. · news |
| LMT | Lockheed Martin Corporation | 606.72 | BUY — QUEUED @567.66 | 567.66 | 549.33 | 613.34 (+8.0%) | 613.68 (+8.1%) | 638.73 (+12.5%) | The strongest buy candidate this book has produced in over a month, and the first in weeks that is neither expensive nor low-quality. Lockheed raised full-year guidance after beating on Q2 - earnings guidance to -30.65 and free cash flow to -7.2bn - on a missile-production surge that pushed its order backlog to a record $230bn, and analysts have moved targets into the -700 range. The valuation screen calls it fairly valued rather than expensive, and unusually for this book the 12-week projection sits BELOW our discounted-cash-flow anchor, meaning the upside does not require heroic growth assumptions. The catch is timing: the shares have already run 6.4% past the model's entry, so we queue a limit order rather than chase it. If it does not pull back within three sessions the order simply expires and we lose nothing. Adj conviction 92 = 98 raw - 8 val_adj (MoS_FV -1.646 -> clamp(round(x5))) + 2 scorecard (PullbackBuy_VWAP|buy, n=22, hit_1w.591 - the book's ONLY positive buy cell). QUALITY GATE PASSES: Quality_Pass True, Val_Signal FAIR_VALUE, Quality_Rank 18, F_Score 8. QUEUED not bought because Entry_Distance_ATR -2.47 (close 606.72 is 2.47 ATR past the 567.66 model entry) with RSI 69.52 - the ACT-NOW no-chase bar is decisive. Provenance clean: 567.66 is the anchor row's printed Entry_Price (Order_Type Buy Limit @); entry range 563.70-571.62 (+-0.25 ATR). Sizing: (85.49/100) x x 20% = x 0.75 modifier (MoS_FV -1.646 in (-3,-1]) = / 567.66 = 45.2 -> =; well inside the 60% x cap (no existing exposure). Stop 549.33 = the model's own Stop_Price (1.16 ATR below the limit, tighter than a 1.5 ATR construction, adopted as-is); risk if filled (1.6% of). NO DCF WARNING - DCF_per_share 692.84 and the 12wk 638.73 sits BELOW it, the first buy card this run whose forward view does not assume growth beyond priced-in. Est_* are computed from the 606.72 close so read high against a 567.66 limit; vs the close they are 1wk +1.09% / 4wk +1.14% / 12wk +5.28%. Target_Price 604.38 is the risk objective, distinct from the 4wk estimate. HONEST CAVEAT: a 6.4%-below-market limit is unlikely to fill inside 3 sessions, and the counterfactual record agrees - no_chase_queue n=15, median wh_1w +1.51% [cost]. That is advisory ONLY, not blended into conviction; the no-chase bar is not relaxed to force a fill (US-20). L1 satisfied: this is a PullbackBuy, not a breakout chase. Regime=Trending. Breaks the ~21-session US-20 buy outage. · news |
| DELL | Dell Technologies Inc | 484.50 | BUY — QUEUED @486.37 | 486.37 | 433.46 | 493.88 (+1.5%) | 519.03 (+6.7%) | 645.97 (+32.8%) | Dell jumped 9.9% to an all-time high after Super Micro guided FY27 revenue to -72bn against a 5bn consensus - a direct read-through to Dell's own AI-server business, which is growing 757% year-on-year against a $43bn backlog and a $60bn FY27 target. The valuation screen still calls it fairly valued with top-decile quality (rank 9, F-score 8), which is rare for anything trading at a record. We queue a stop just above the 52-week high so the order only fills if the breakout is real, and we size it three-quarters of normal because the stock is no longer cheap. The honest counterweight: UBS argues the shares have little left after roughly doubling this year, and this signal family has yet to hit a 4-week target in seven attempts. Adj conviction 68 = 78 raw - 10 val_adj (MoS_FV -2.881) + 0 scorecard (BreakoutUp_52W|buy, n=7, hit_1w.50, hit_4w.00 - actionable and NEUTRAL, unlike every other buy cell in the pool). QUALITY GATE PASSES: Quality_Pass True, FAIR_VALUE, Quality_Rank 9, F_Score 8. Clears the 65 floor by 3. L6 CHECKED AND CLEARED - this is NOT a post-earnings gap: the move is an SMCI peer-guide read-through, Dell's own FQ2 is still ahead. L6 bars chasing a name's OWN earnings gap; a peer re-rate is not that. Provenance clean: 486.37 is the anchor row's printed Entry_Price (Order_Type Buy Stop @), Entry_Distance_ATR +0.05, trigger above the 484.50 close and just above High_52W 485.70 - self-limiting. Entry range 477.55-495.19 (+-0.25 ATR). Sizing: x 0.75 = / 486.37 = 52.7 -> = STOP 433.46 (1.5 ATR below trigger); the model's Stop_Price 376.66 is REJECTED as far too loose - 22.6% below entry is not a signal-defined risk line by this book's standard (US-29 applied to a new entry); risk if filled WARNING: 12wk 645.97 exceeds the 485.61 DCF anchor (1.33x) - projection assumes growth beyond priced-in; mildest overshoot in the book (KLAC 2.55x, GOOGL 2.64x). Target_Price 705.80 is the risk objective. CONDITIONAL ON THE 08-05 RECONCILE - THIS CARD DOES NOT FIRE WHILE THE BOOK IS UNRECONCILED ON DELL: state carries DELL DISPOSITION OWED from the 08-05 BUY 55 @464.24 while input/usportfolio.json shows (session 6). If that fill is recorded TAKEN this card is WITHDRAWN, not doubled - you would already own at 464.24 and adding 52 more at 486.37 is paying up 4.8% on a cell that is 0-for-7 at 4 weeks. If recorded SKIPPED the queue stands as written. Sizing assumes the file's is ground truth. Run 'trade buy' or 'trade skip us DELL' to adjudicate. · news |
| NET | Cloudflare, Inc. | 311.45 | HOLD | 311.45 | 282.89 | 310.98 (-0.2%) | 321.69 (+3.3%) | 337.45 (+8.3%) | Holding Cloudflare and declining to add to it for a third session. The signal strengthened sharply - conviction - and on the arithmetic alone it now clears our bar. But we committed in writing that we would only add on a pullback to the 20-day average, and the stock is 9.2% above that line. Buying here is the chase the rule exists to prevent, especially on a signal family that has hit its 1-week target only once in seven tries. We raise the trailing stop instead, from 279.08 to 282.89. Adj 67 = 84 - 10 val_adj (MoS_FV -943.71, an EPV artifact on a name whose FV_per_share is) - 7 scorecard (FailedBreakdown_Up_20D|buy, n=7, hit_1w.143 - the book's worst actionable buy cell). Two points ABOVE the floor, recovered from 51 on 08-12. THE FLOOR IS NECESSARY, NOT SUFFICIENT: the re-arm pre-committed 08-11 and re-based 08-12 requires BOTH a pullback to the live EMA20 +-0.25 ATR AND conviction >=65; only the conviction leg is met (EMA20 285.29, close 311.45 = +9.2% above). L1 reinforces: favour PullbackBuy over FailedBreakdown chases. RE-ARM CARRIED, RE-BASED to EMA20 285.29 +-0.25 ATR (280.53-290.05) with conviction >=65. Trail 1.5 ATR = 282.89 above the standing 279.08 -> RAISE, new Peak Stop; the model's 301.93 at 0.50 ATR is tighter, noted, NOT adopted (whipsaw). DATA DEFECT, 13TH SESSION: two NET rows in input/usportfolio.json (150 @258.00 with Name misspelled 'Clouadflare Inc' / 0); the loader takes the last, so this position reads Held=N to the scan (no Avg_Cost/Gain_ATR/T12_Progress, no ADR-0012 arm possible) and unledgered_drift to the monitor every run. DE-DUPLICATE. |
| NVDA | NVIDIA Corporation | 224.09 | HOLD | 224.09 | 213.10 | 225.75 (+0.7%) | 229.51 (+2.4%) | 241.14 (+7.6%) | Holding Nvidia and raising the trailing stop from 212.52 to 213.10. The stock rose 3% to 224.09 and sits above every moving average we track, but a momentum-divergence warning has appeared alongside the profit-armed flag - the kind of signal that would normally justify banking a slice of the gain. We cannot act on it, because the portfolio file has no recorded for this position, so the system cannot verify the gain is real. This is the third consecutive session a valid profit-take has been lost to that one blank field. Earnings 26-Aug. NO TRADE cv0 -> HOLD cv93 on AnticipationUp_VCP; Exit_Warning now prints RSI_DIV,PROFIT_ARMED = ADR-0012 tick (a) on an armed name (T12_Progress 92.93 >= 90), squarely. The G1 softening does NOT rescue it either way: G1 restricts tick (a) only on a fresh BreakoutUp_*/MomentumContinuation_* at conviction >=80, and AnticipationUp_VCP is NEITHER family, so (a) would fire cleanly. US-26 DECLINES THE ARM OUTRIGHT: Avg_Cost blank in input/usportfolio.json so Gain_ATR and Unreal_PnL_Pct are blank and the gain is unverifiable - no armed position exists for (a) to fire against, and a bare HOLD cannot originate a sell (US-17). Close 224.09 above EMA10 215.31 / EMA20 211.01 / EMA50 207.39 / EMA200 195.87, RSI 62.42. Trail 1.5 ATR = 213.10 above the standing 212.52 -> RAISE. ADR-0020: Peak Stop 218.23 (07-09) vs 213.10 = -2.35%, no fire. SHARED WITH ustech/ussemicon - execute once (L7). ACTION REQUIRED: fill Cost for NVDA in input/usportfolio.json - highest-value unshipped fix in the book (systemic #2). |
| MU | Micron Technology, Inc. | 911.29 | HOLD | 911.29 | 815.47 | 916.73 (+0.6%) | 1009.11 (+10.7%) | 1136.10 (+24.7%) | Holding Micron and taking the biggest trailing-stop raise of this run, from 774.61 to 815.47. The stock rose 4.9% to 911.29 in the semiconductor rally that followed the cooler July inflation print, moving back above its 20- and 50-day averages and 6% clear of our cost. The signal engine still declines to rate it - eleven sessions of NO TRADE on consensus conflict - but the 12-week view remains 25% higher and nothing in the risk framework calls for a cut. Applied Materials reports after the close on 13-Aug and is the next read-through for the group. NoTrade_Reason verbatim: 'Consensus conflict + low conviction.' 11th consecutive NO TRADE cv0. +6.03% over the 859.49 cost; above EMA20 891.78 / EMA50 888.54, far above EMA200 616.90. Target_12Week 1136.10 = +24.7%, alive. NOT ARMED (Gain_ATR 0.81, T12_Progress 80.21). ADR-0020 RE-CHECKED: Peak Stop 880.00 (07-22) vs live 774.61 = -11.98%, INSIDE the 15% threshold - no fire; tonight's raise to 815.47 lifts the line to -7.33%, further inside. Trail 1.5 ATR = 815.47 above the standing 774.61 -> RAISE. Quality_Pass True, Quality_Rank 1 (best in the universe), F_Score 8. · news |
| LLY | Eli Lilly & Co | 1220.28 | HOLD | 1220.28 | 1159.23 | 1232.18 (+1.0%) | 1248.24 (+2.3%) | 1293.44 (+6.0%) | Holding Eli Lilly with the stop unchanged at 1159.23. Conviction recovered from, the stock edged up to 1220.28 and sits above its 20-, 50- and 200-day averages with a 12-week view 6% higher. Nothing here argues for trimming a winner, and the valuation screen cannot be used to argue otherwise because its cash-flow anchor is more than three times below the market price - unreliable by our own rule. HOLD cv62 -> cv68 on FailedBreakdown_Up_20D, +0.43%. PROFIT_ARMED with T12_Progress 94.34, but US-26 DECLINES THE ARM (Avg_Cost blank in input/usportfolio.json, so Gain_ATR and Unreal_PnL_Pct are blank and the gain is unverifiable). Even with a valid arm nothing fires: (a) only PROFIT_ARMED, (b) conviction ROSE 6 pts, (c) Target_12Week 1293.44 = +6.0% over the close, alive, (d) not a NO TRADE. G3 would independently block a valuation-amplified trim (bullish HOLD, positive 12wk, above EMA20 1183.70 / EMA50 1150.93 / EMA200 1032.19, RSI 58.08 - cutting the winner) and G2 disqualifies the amplify anyway (DCF_per_share 387.04 vs a 1220.28 close = 3.15x overshoot, so MoS_FV -3.36 cannot bump a tier). Trail 1.5 ATR = 1156.29, BELOW the standing line -> no raise, stop STANDS. ADR-0020: Peak Stop 1180.30 (07-03) vs 1159.23 = -1.79%, no fire. ACTION REQUIRED: fill Cost for LLY in input/usportfolio.json (systemic #2). |
| OKTA | Okta Inc. | 147.30 | HOLD | 147.30 | 141.21 | 147.78 (+0.3%) | 155.01 (+5.2%) | 167.02 (+13.4%) | Holding Okta with the stop unchanged at 141.21. The stock slipped 2% to 147.30 and, for the first time in five sessions, is no longer close enough to its 12-week target to be flagged for profit-taking at all - so the question of trimming does not arise. The position is 5.4% ahead of cost and the 12-week view is still 13% higher. Earnings 26-Aug. HOLD cv68 -> cv61 on FailedBreakdown_Up_20D. DE-ARMED: T12_Progress fell 91.48 -> 88.19, below the 90 arm threshold, and Gain_ATR 1.22 is nowhere near 3.0, so PROFIT_ARMED no longer prints and there is no armed position at all. The -7 conviction drop is well short of tick (b)'s -20 bar and moot without an arm. Trail 1.5 ATR = 138.09, BELOW the standing line -> no raise, stop STANDS. ADR-0020: Peak Stop 141.21 (08-12) vs 141.21 = 0%, no fire. Close 147.30 above EMA10 146.26 / EMA20 143.62 / EMA50 132.50 / EMA200 105.61, RSI 57.77. Quality_Pass False but irrelevant - the gate blocks new entries only, never held names. |
Outcome & track record
Accuracy at the time of this record.
127 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 →
-0.148 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.