Recommendations
3 to acquire · 2 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 | 205.76 | SELL 30% | 205.76 | 193.38 | 207.15 (+0.7%) | 213.30 (+3.7%) | 214.31 (+4.2%) | Trimming 30% of KLA on risk discipline alone, not on any view about the business. The protective stop on this holding has drifted 28.9% below its own high-water mark (272.00 down to 193.38) while the position has never once been reduced in its entire history, and a risk line walking that far away from an uncut holding is a risk failure in itself. The company is in good shape - the quarterly dividend was just declared at and AI fab demand is being reinforced by projects like the Tesla/SpaceX Terafab complex - but the shares sit roughly 3.5x above their intrinsic-value anchor, the stock is down about 9% over the past month on margin and geopolitical worries, and the model's own twelve-week view is only +4.2% from here. Very little upside is given up for a meaningful cut in exposure; stay long the thesis. OPS: ADR-0020 first fire, SESSION 14 unexecuted - carded 07-28 through 08-15 without a share moving; US-37 stalemate, requires an explicit TAKEN/SKIPPED disposition. Tier fixed at the 30% floor. US-36 veto-immune: US-17 (bare NO TRADE), US-18/34 (lone-flip bar), US-22 and US-30 DO NOT APPLY to this first fire - without that immunity the card dies tonight because the row is NO TRADE cv0. Signal deteriorated -> across four snapshots (NoTrade_Reason verbatim: 'Consensus conflict + low conviction') but that is not the basis. ADR-0012 checked and DECLINED: now armed on BOTH limbs (Gain_ATR 3.16 exceeds 3.0 for the first time so US-33 no longer binds; T12_Progress 96.01; US-26 passes at +17.08% unrealised) but no tick qualifies - (a) no 2nd Exit_Warning code, (b) is a label flip not a within-label decay, (c) T12 214.31 = 1.0416x close is alive, (d) first NO TRADE not the second. Valuation amplify to 50% SUPPRESSED by G2 (T12 214.31 vs DCF 61.81 = 3.47x overshoot, worsening from 2.60x/3.32x). Trail: 1.5 ATR = 191.50 is looser, so stop 193.38 STANDS; Entry_Price blank on a NO TRADE row so the settled close is the action reference. Re-price of an unexecuted card (203.72 -> 205.76, worth +on the delay), not a US-19 escalation. Realised gain approx +vs the 175.75 cost. · news |
| GOOGL | Alphabet, Inc.- Class A | 344.00 | SELL 30% | 344.00 | 332.73 | 343.04 (-0.3%) | 339.95 (-1.2%) | 378.63 (+10.1%) | Banking a third of a position that is up 180% from cost, into a stalling tape. Alphabet has now closed three straight sessions with the model finding no tradable setup at all, the shares have slipped below their 20-day average, and the stock fell this week on a shakeup in the AI division including the departure of Jeff Dean - a live question mark over the exact franchise this position is held for. The counterweight is real: Berkshire Hathaway raised its stake 83% in the second quarter to about 106 million shares (~$37bn, now its third-largest holding), including a $10bn direct purchase in Alphabet's June $85bn AI-infrastructure raise. But that describes June, while three flat closes and a leadership exit describe now, and the model's own one-week and four-week views are slightly NEGATIVE from here against a +10% twelve-week view. Take a third off, ride into the longer view. OPS: ADR-0012 profit-take, armed on the Gain_ATR limb (27.32 vs a 3.0 bar, Avg_Cost 123.00, Unreal_PnL +179.67%) so US-26 passes cleanly and US-33 does NOT bind. Tick (d) fires on three consecutive settled NO TRADE cv0 prints (A13C/A14C/A15C), deepened from the two that fired it on 08-15. One soft tick = profit_take_1 = 30%; tier deliberately NOT escalated - US-19 is satisfied by executing this first fire. Ticks (a)/(b)/(c) decline: PROFIT_ARMED only, T12 378.63 = 1.1006x close alive. SESSION 5 unexecuted; US-37 stalemate, requires TAKEN/SKIPPED. The overlay's 'close below 345' line and the 08-08 pre-committed 350.66 line are BOTH SET ASIDE - the overlay is 44 days stale (US-31 names this exact line as the one that drove six unexecuted sessions before healing) and US-27 requires re-verification; the card rests on ADR-0012 alone. Valuation val_adj -3 too shallow to amplify and G2 would suppress anyway (378.63 vs DCF 145.29 = 2.61x). Trail: 1.5 ATR = 331.87 is looser, stop 332.73 STANDS. Entry_Price blank on a NO TRADE row so the settled close is the action reference. Re-price 345.90 -> 344.00 (-on the delay). Realised gain approx +vs the 123.00 cost. · news |
| ACN | Accenture PLC | 169.98 | BUY — QUEUED @164.84 | 164.84 | 156.23 | 171.22 (+3.9%) | 167.46 (+1.6%) | 173.09 (+5.0%) | A pullback order on the highest-quality new name the screen has produced in weeks. Accenture has fallen more than 50% from its January level on a genuine fear - that agentic AI automates a large share of what a 700,000-person consulting workforce does - but the business underneath is not breaking: last quarter revenue rose 6% to 7bn with 6bn of free cash flow, 2bn returned to shareholders, and record bookings of 1bn including 41 clients above, with agentic-AI partnerships now running with OpenAI and with Microsoft/Avanade. The stock has turned up about 22% over the past month, which is the recovery this order is buying, and it is the only name on the board that is both fairly valued and passes the quality screen outright. The order sits 3.1% below the market on purpose: the disruption thesis is real enough that this is worth owning on a pullback, not at the price. OPS: BREAKS the 14-session US-20 buy outage - Quality_Pass TRUE (F 6, Quality_Rank) is the differentiator; every PullbackBuy candidate since mid-July was quality-gate-killed. Buy_Rank. Trail NO TRADE -> WATCH 58 -> BUY 67 -> BUY 74, three strengthening prints, fresh and stable. Adj 74 -1 (val) +2 (PullbackBuy_VWAP|buy cell, the sole positive buy cell) = 75. QUEUED not market: Entry_Distance_ATR -0.69 is past the -0.25 no-chase bar. Entry range 162.98-166.70 (+-0.25 ATR). Risk /sh = 5.2% =; Target_Price 182.07 gives R:R 2.0. DCF_per_share 370.35 vs a 176.89 valuation price is the widest DCF discount in the universe - no overshoot annotation applies, the 12wk estimate sits far BELOW the DCF anchor. AUTO-WITHDRAW on conviction <65 or a family flip off PullbackBuy_VWAP (US-8); expires 2026-08-20. Note US-24: the PullbackBuy edge is 1wk-only and the cell has eroded +5 -> +3 -> +2 (hit_1w.591, hit_4w.25) - treat Target_Price as the risk objective, not an expectation. · news |
| VIST | Vista Energy, S.A.B. de C.V. | 72.16 | BUY-STOP — QUEUED @71.44 | 71.44 | 62.34 | 72.53 (+1.5%) | 73.15 (+2.4%) | 77.36 (+8.3%) | The single cheapest quality name on the board, bought only if it keeps going up. Vista Energy - one of the largest oil producers in Argentina's Vaca Muerta shale - is the only holding candidate the valuation screen calls outright undervalued while also passing the quality test, with an intrinsic-value anchor around against a share price near Two things have just changed in its favour: Peter Thiel's Thiel Macro disclosed a roughly 1% stake, about 1.2m shares for, now the fund's second-largest position behind only Amazon, and the shares jumped 5.4% on the filing; and the company itself has been buying back stock through the period, including on the decision date. The order is a buy-stop just above the market, so it only fills if the move continues - if the Thiel pop fades, nothing is bought. Sized at half a normal position because the risk line is wide (12.7%) and the company did just miss on earnings. OPS: ANTICIPATE slot. Eligibility met - BUY - ANTICIPATE in 2 of the last 3 snapshots (A15I cv75, A15C cv75), conviction 75 >= the 55 floor. THE 08-15 PRE-COMMITMENT FIRES AS WRITTEN: that entry carded VIST at adj 62 and pre-committed to acquire on a PullbackBuy print >=65 OR any family raw >=75 - tonight prints raw 75, the second limb exactly (US-35). Adj 75 +2 (val, full credit on Quality_Pass True) +0 (AnticipationUp_VCP|buy is n=1, REPORT-ONLY, no bias applied) = 77, the highest in the book. Size: 60.76% x x 20% = x1.25 (UNDERVALUED + Quality_Pass) = x0.5 (anticipate half-size) = / 71.44 =. Entry_Distance_ATR -0.24, just inside the -0.25 no-chase bar. Risk /sh = 12.7% =; Target_Price 89.64 gives R:R 2.0. The 12wk estimate 77.36 sits BELOW DCF_per_share 86.44 - no overshoot annotation, the projection stays inside the intrinsic anchor, the opposite of every held name in this book. COUNTERFACTUAL ARGUES AGAINST AND IS ANSWERED: counterfactuals.anticipate n=6 median wh_4w -9.86 [saved] says non-fills have been saving money, contradicting US-21; per US-28 that aggregate is regime-dependent and advisory only and is NEVER blended into conviction - VIST is not the marginal name it was built from (only UNDERVALUED + Quality_Pass True row, named institutional catalyst, live buyback, 12wk inside DCF, already half-sized). Expires 2026-08-20. · news |
| NFLX | Netflix, Inc | 76.02 | BUY — QUEUED @72.99 | 72.99 | 70.27 | 76.11 (+4.3%) | 74.45 (+2.0%) | 74.85 (+2.5%) | A second pullback order on a quality name, deliberately kept small. Netflix has firmed up over the last two sessions into the model's second-best buy setup, and it passes the quality screen, but it is priced well above its intrinsic-value anchor and - stated plainly - the model's own four-week and twelve-week views both sit BELOW where the stock trades today. That means the trade only works because the order is placed 4% under the market: from the order price the twelve-week view is +2.5%, which is a thin expectation. Hence a three-quarter-size order and second billing behind Accenture. If the pullback does not come, nothing is bought and nothing is lost. OPS: PullbackBuy_VWAP, Buy_Rank, trail NO TRADE -> NO TRADE -> BUY 58 -> BUY 72 (two prints, fresh). Adj 72 -7 (val, MoS_FV -1.469) +2 (cell) = 67, clears the 65 floor by two. Quality_Pass TRUE (F 7, Quality_Rank - second-best quality in tonight's pool). Size modifier 0.75x (-3 < MoS <= -1): 49.24% x x 20% = x0.75 = / 72.99 =. QUEUED not market: Entry_Distance_ATR -1.29, the deepest extension in the pool, limit 4.15% below the close. Entry range 72.40-73.58. Risk /sh = 3.7% =; Target_Price 78.43 gives R:R 2.0. WARNING - Est_4W 74.45 and Est_12W 74.85 are both BELOW the 76.02 close: 8th appearance of the bullish-row/sub-close-12wk defect (systemic #4), logged and NOT smoothed over. NEWS NOT SWEPT - the 6-name shortlist cap was hit (KLAC, GOOGL, ACN, VIST, OKTA, SPCX) and NFLX was dropped as the lowest-urgency card; RE-CHECK AT THE 08-18 BOD BEFORE ANY FILL. Auto-withdraw on conviction <65 or a family flip (US-8); expires 2026-08-20. |
| MU | Micron Technology, Inc. | 1011.75 | HOLD | — | 921.42 | 1021.85 | 1051.00 | 1174.44 | The standout of the session: Micron gained 4.1% to its first four-figure close and is now up 17.7% from cost, with the model's conviction strengthening for a third straight reading. Raising the protective stop to 921.42 locks in a 7% gain over cost while leaving the position fully exposed to a twelve-week view still 16% above here. Nothing to do but let it run behind a higher line. OPS: HOLD cv69 ->, the highest conviction in the book, on AnticipationUp_VCP. Above EMA20 914.61, far above EMA200 627.34. NOT armed - Gain_ATR 2.53 < 3.0, T12_Progress 86.15 < 90, no Exit_Warning - so the ADR-0012 profit-take path is correctly unavailable. TRAIL RAISE 882.92 -> 921.42 (1.5 ATR = 90.33), the largest raise of the run; this sets a NEW Peak Stop dated 2026-08-17 and keeps ADR-0020 erosion at 0%. Quality_Pass True, Quality_Rank, F 8 - the best quality in the universe. HOLD is not a buy action so MU does not enter the buy pool despite cv81. SHARED WITH ustech/ussemicon - execute once (L7). |
| NBIS | Nebius Group, N.V. | 268.85 | HOLD | — | 235.69 | 271.10 | 291.81 | 310.12 | The de-risk on Nebius is done - were sold at on 16 August, slightly more than the 90 that were carded and at a good price - so the position is back to a clean hold on, still up 27% from cost with a twelve-week view 15% above here. The stop stays at 235.69, which locks in a gain, and no further action is warranted. OPS: ADR-0020 first fire DISCHARGED; Peak Stop watermark RE-BASES 270.00 -> 235.69, erosion resets to 0%, and the first-fire veto-immunity is now SPENT - any future NBIS fire re-enters the full veto stack. Tonight HOLD cv54 -> cv61, recovering from the intraday but a -27 point two-session decline from the cv88 peak of 08-14; that would be ADR-0012 tick (b) IF the position were armed, which it is NOT (Gain_ATR 2.22 < 3.0, T12_Progress 86.69 < 90, no Exit_Warning) - so no profit-take path exists and none is claimed. -3.18% on the session after +8.88%. MODEL Stop_Price 255.95 REJECTED as a trail: only 0.50 ATR below the close on a name whose ATR is 9.6% of price, it would be stopped out by an ordinary session (systemic #5, the tight-print variant of US-29). Book convention 1.5 ATR = 230.15 sits below the standing line, so stop 235.69 STANDS (+14.1% room). SHARED WITH ustech - execute once (L7). |
| OKTA | Okta Inc. | 143.24 | HOLD | 144.56 | 136.11 | 143.70 (-0.6%) | 154.52 (+6.9%) | 166.58 (+15.2%) | Okta slipped 2.8% and closed 1.1% under its protective stop - but this is not a sell and not an add. It is a marginal breach, the model simultaneously flipped the name to its most reliable buy setup for the first time in this run, and the twelve-week view is +16% - the opposite of the deteriorating-forward condition that would make a stop break an exit. Citizens has just upgraded the stock to Market Outperform with a target and the company closed a acquisition of Permiso Security for identity threat detection; shares are up 70% year to date. The stop is reset to 136.11 and the position rides into the 26 August earnings print. OPS: THE MOST CONTESTED CALL TONIGHT. Sell DECLINED on four independent grounds - (i) L2 requires stop-break AND a negative 12wk, and T12 166.58 is +16.3%; (ii) US-22 marginal-breach clause, 1.10% is not decisive; (iii) US-30, a stabilization print outranks a break and HOLD cv57 -> BUY - ADD cv71 on PullbackBuy_VWAP with Buy_Rank is exactly that; (iv) news confirms. ADD ALSO DECLINED: raw 71 + val_adj -10 (MoS_FV -5.008, floored) + bias +2 = ADJ 63, two points below the 65 floor - and independently Est_1W 143.70 is flat to the 144.56 limit while earnings land 08-26 with a horizon stance of TRIM-INTO and a +-10% band, which is not a week to add size. Not armed (Gain_ATR 0.47, T12_Progress 85.99, no warning code). STOP RESET 144.84 -> 136.11 (the fresh PullbackBuy risk line, above the 132.30 chandelier). WARNING: this reset is step one of a NEW ADR-0020 erosion clock at -6.03% - the exact path KLAC walked across 20 sessions, every breach answered by lowering the line, size never cut). Two more resets of this size fire ADR-0020 on OKTA. · news |
| NVDA | NVIDIA Corporation | 225.01 | HOLD | — | 217.07 | 226.45 | 229.50 | 239.47 | Nvidia is held into its 26 August earnings behind a slightly higher stop at 217.07. The model has now gone two straight sessions finding no tradable setup, which on a position this profitable would normally trigger a partial profit-take - but the for this holding is missing from the records, so the gain cannot be verified and the trim rule cannot be applied by rule rather than by guess. Fixing that record is the single highest-value data task in this book. OPS: HOLD cv59 -> NO TRADE cv0 (NoTrade_Reason verbatim: 'Consensus conflict + low conviction'). This is the SEVENTH consecutive session the ADR-0012 arm has been declined. Tonight tick (d) IS fully satisfied - two consecutive settled-close NO TRADE prints (A14C and A15C) - and the T12-limb arm (T12_Progress 93.96) has the non-(c) tick US-33 demands. It still cannot fire because US-26 declines the arm outright: Avg_Cost is BLANK in input/usportfolio.json so Gain_ATR and Unreal_PnL_Pct are both blank and the gain is unverifiable; decisions/us_transactions.jsonl cannot supply it either because NVDA enters as a 07-04 'adjust' of with price: null. Quantified cost of the missing field: the declined 30% tier is Above EMA20 214.68 / EMA200 196.81, Regime Transitional. TRAIL RAISE 216.38 -> 217.07 (1.5 ATR = 7.94); peak 218.23 unchanged, erosion -0.53%. Earnings 08-26, 7 sessions out, hold-through, +-6.75% band. SHARED WITH ustech/ussemicon - execute once (L7). FILL Cost FOR NVDA - systemic #1. |
| LLY | Eli Lilly & Co | 1183.16 | HOLD | — | 1159.23 | 1193.26 | 1211.89 | 1262.43 | Eli Lilly has stabilised - the model found a tradable hold setup for two straight readings after two sessions of nothing - so the partial profit-take that was hanging over this position has dissolved on its own terms and is dropped rather than carried forward. The stop stays at 1159.23, which is now the tightest line in the book at 2.1% below the market, with a twelve-week view 6.7% above here. OPS: NO TRADE cv0 x2 -> HOLD cv54 x2 (A15I and A15C) on FailedBreakdown_Up_20D. The 08-15 entry recorded a fully-formed tick (d) surrendered to a blank Avg_Cost; tonight the label itself recovered so the tick is GONE on its own terms - DE-ESCALATED per US-32, which is the correct mirror of US-19's no-escalation rule. PROFIT_ARMED still prints with T12_Progress 93.72 and Avg_Cost is STILL blank (07-04 'adjust' of, price: null), so the arm remains unverifiable under US-26 regardless. +0.25% to 1183.16, marginally below EMA20 1185.32, well above EMA50/EMA200 1037.65. MODEL Stop_Price 1104.64 REJECTED - it sits 4.7% BELOW the standing line, the US-29 silent-loosening artifact (systemic #5); 1.5 ATR = 1124.27 is also below. Stop 1159.23 STANDS, +2.06% room. FILL Cost FOR LLY - systemic #1. |
| NET | Cloudflare Inc | 307.01 | HOLD | — | 301.75 | 307.79 | 311.97 | 331.92 | Cloudflare fell 2.8% and the model has now found no tradable setup for three straight sessions, so there is nothing to act on - but the stop at 301.75 is now only 1.7% below the market on a position, which makes this the line most likely to be tested next. The position is up 12% blended; the bought at on 16 August, outside any recommendation, are already 5% underwater. Hold and watch the line. OPS: NO TRADE cv0 x3 (NoTrade_Reason verbatim: 'Consensus conflict + low conviction'). No buy row exists so the ADD arithmetic is moot; the 08-14 re-arm is NOT triggered and is CARRIED, RE-BASED to the live EMA20 293.54 +-0.25 ATR (289.18 - 297.90) with conviction >= 65. The 08-16 buy of 50 @323.00 executed outside any card - recorded, not criticised, but it is an unmodelled position; blended cost 274.25 on = +11.95% while the new tranche alone is -4.95%. Quality_Pass False (F 4, Quality_Rank 36, MoS_FV -9.70) - irrelevant to a hold, but it is why no NET buy has ever been carded. Trail: 1.5 ATR = 280.84 is looser, stop 301.75 STANDS (+1.71% room, second-tightest in the book after LLY). WARNING, 15TH SESSION: two NET rows in input/usportfolio.json (200 / 0); the loader takes the last, so this position reads Held=N to the scan and triggers a false unledgered_drift warning every night. DE-DUPLICATE - systemic #8. |
| SPCX | Space Exploration Tech Corp | — | HOLD | — | 129.00 | n/a | n/a | n/a | SpaceX passed its first and biggest lock-up test convincingly - roughly 911m insider shares became sellable on 6 August and the stock rose 6.1% that day, then rallied hard across the following week, adding several hundred billion in market value and reclaiming its IPO price, with Morgan Stanley calling the fundamentals largely unchanged and current levels an attractive entry. The next 7% tranche unlocks on 21 August; the position holds through it on the existing stop, with no trim and no further adds into the event. OPS: PERMANENTLY MUTED on signals (no_close, private-market row, absent from the scan) - qualitative read only, nothing synthesized. Holding is now after a 140 @139.89 buy on 08-16 that executed OUTSIDE any card, marked at the last observed public print = the book's third-largest position. Classification: CONFIRMS-HOLD. HORIZON: lock-up tranche 2 (~7%, up to ~319m shares) lands 08-21, 4 sessions out, band -8/+5%, stance hold-through - INSIDE the 5-session window, so the catalyst-aware hold formally applies: no trim into the unlock, no further adds. Stop 129.00 STANDS; re-arm on a settled close below 129.00. AUTO-UNMUTE the moment SPCX appears in the scan CSV with a non-null Close. TWO DATA DEFECTS: (1) Cost reads 65.282 = 19,584.60/300, i.e. ONLY the new tranche's cash spread over the whole position - the original ' basis has been OVERWRITTEN and is unrecoverable from the ledger (07-04 adjust, price: null); any or ADR-0012 arm on this position is now wrong by construction (systemic #2). (2) The mark may materially understate the position - the reported ~35% post-lockup rally from implies something closer to a potential understatement on a holding (systemic #3). · news |
Outcome & track record
Accuracy at the time of this record.
137 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 →
-0.132 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.