US — End-of-Day Verdict
ACQUIRE NO ACTION (0 of <=3) - 15th consecutive session with no admissible new BUY (US-20): QCOM adj 66 was the sole floor-clearer AND quality-passing (F=5, ROIC 22.99%, FAIR_VALUE) but was rejected on reward:risk - a -16.6% stop against modelled forwards of +0.22%/+1.29%/+0.98% (~0.06:1), below all four EMAs, Consolidation_Bearish, on a signal family with zero graded history; DDOG adj 69 CLEARED the floor at cv86 Buy_Rank 1 and was killed by the hard quality gate (Quality_Pass False, F=5, ROIC 2.49%); CRM adj 59 (below EMA200 198.59); BAC adj 59 (Quality_Pass False); SNOW adj 50 (Quality_Pass False, RSI 74.5, extended). Zero PullbackBuy_VWAP candidates for a 4th session. ANTICIPATE 0 of <=1 - the US-35 pre-committed VIST/WDS trigger did not arm: both printed BUY - ANTICIPATE on 08-01 and both went to NO TRADE cv0 with Conflicted pillars at the anchor. | DISPOSE 2 of <=3: KLAC 50% trim @ ~182.75 - 3rd consecutive STRONG SELL - REDUCE with the only label-agreeing bearish pillars in the book, AND ADR-0020 TRAIL EROSION first fire (-38.24% vs a 272.00 Peak Stop, zero size reduction), which makes it mandatory rather than permitted; G2 suppresses the MoS -4.41 tier bump; keep-core 500. NVDA 30% trim @ ~206.64 - Sell_Rank 1, adj 79 (highest on either side), forward dead at all three horizons (-0.04%/-1.60%/+0.35%), lone-flip bar overridden on US-18's own 'live forward view' clause; execute ONCE vs ussemicon. GOOGL trim DEMOTED to a trail-raise on a news override; NBIS trim DECLINED as a lone flip on the stabilization session. | The session was risk-on with the laggards leading: NBIS +11.63% ( GPU-backed debt facility, Nasdaq-100 inclusion nearing), GOOGL +4.88% (Morgan Stanley AI-capex reassurance), NVDA +2.93%, against LLY -2.39% into its 08-05 print. | : buys sells free -> (~30.5% cash on a mark).
Recommendations
0 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 | 182.75 | SELL 50% | 182.75 | 168.00 | 184.15 (+0.8%) | 187.60 (+2.7%) | 190.45 (+4.2%) | Halving the book's largest position after a third straight session of the model printing a strong sell on KLA, and after the protective stop was walked down 38% while the position was never once reduced. The technical picture is the only unambiguously bearish one we hold: price is below its 20-, 50- and 100-day averages, RSI 38, and the stock sits 40% below its 52-week high. Fair value agrees the shares are dear (margin of safety -4.4, growth already priced in), while the business itself is doing well - record quarterly revenue, a raised outlook and management calling demand 'accelerating' - which is exactly why we are halving rather than exiting, keeping for the recovery. OPERATIONAL: US-18/34 satisfied 2nd session (3 consecutive settled STRONG SELL - REDUCE cv76/76/68, Sell_Rank 3, pillars Consolidation_Bearish - the only label-agreeing bearish structure in the book); ADR-0020/US-36 TRAIL EROSION FIRST FIRE - live stop 168.00 vs Peak Stop 272.00 (2026-07-01) = -38.24%, zero size reduction since, veto-immune, converts the card from permitted to mandatory (30% floor; the stable-REDUCE basis carries the standing 50%). G2 SUPPRESSES the MoS<=-3 tier bump (Est_12W 190.45 = 1.91x DCF anchor 99.91). Sized 50% not the engine's 88.52% per US-12/US-22 keep-core; re-fire of the 08-01 card at the same tier, NOT an escalation (US-19). Position is 26.3% of a mark -> 13.1% post-trim. Retained 500 keep the 168.00 trail (no double-tightening into a 13.57 ATR); Peak-Stop watermark RE-BASES to 168.00 on execution, so any future ADR-0020 fire is subject to the full veto stack again. AMD Q2 08-04 AMC is an ecosystem read-through only - catalyst-aware hold does not attach. US-37 STALEMATE: 4th session unexecuted (07-28 30%, 07-29 50%, 08-01 50%) - requires an explicit TAKEN or SKIPPED disposition. WARNING watermark hygiene: KLAC's pre-split stops (1994.40/1833.36, before the 10-for-1 split effective 2026-06-12) are unadjusted in the ledger and were excluded by hand - see systemic finding 1. · news |
| NVDA | NVIDIA Corporation | 206.64 | SELL 30% | 206.64 | 194.00 | 206.56 (-0.0%) | 203.33 (-1.6%) | 207.37 (+0.4%) | Taking 30% off Nvidia because our model now prices essentially no return from here over any horizon we measure - one week -0.04%, four weeks -1.6%, twelve weeks +0.35% - while the stock carries the strongest sell reading in the portfolio today. The shares rose 2.9% on Monday and AMD's results tomorrow evening could set a friendly tone for the whole AI-chip complex, which is precisely why we are trimming less than a third and leaving exposed to that upside. Valuation supports the caution without driving it: growth is already priced in, and the twelve-week projection sits well above any defensible cash-flow anchor. OPERATIONAL: Sell_Rank 1, adj 79 = highest adjusted conviction on either side of the book tonight, on the best-graded sell cell (FailedBreakout_Down_20D|sell +3, n=13, hit_1w.636). US-18/34's lone-flip bar (J31C was NO TRADE cv0) is OVERRIDDEN on the rule's own text - it protects 'an armed held winner WITH A LIVE FORWARD VIEW', and NVDA has no live forward at any horizon; the same discriminator keeps GOOGL (+8.54% 12wk) and NBIS (+43.2%) protected tonight. The 08-01 basis (ADR-0012 armed via T12_Progress + soft tick (c)) is FORMALLY RETIRED, not carried: US-33 bars tick-(c)-alone on a T12-armed name, and US-26 blocks it independently because NVDA has NO Cost in usportfolio.json (Gain_ATR/Unreal_PnL_Pct null - the position cannot be verified as a winner). This card stands on the signal or not at all. MoS -2.07 > -3 so no tier bump; G2 would suppress anyway (Est_12W 207.37 vs DCF 163.98). SHARED WITH ussemicon - EXECUTE ONCE (L7). US-37 STALEMATE: an unexecuted NVDA trim has been standing since 07-16 (13+ sessions) - requires an explicit TAKEN or SKIPPED disposition; if the standing human answer is SKIP, record SKIP and the card stops being written. Retained 70 raise the trail 188.00 -> 194.00 (~1.5 ATR below close, just above EMA200 193.85), which also widens the ADR-0020 watermark gap from -13.85% to -11.1%. · news |
| GOOGL | Alphabet, Inc.- Class A | 373.51 | HOLD | 373.51 | 343.50 | 378.23 (+1.3%) | 382.46 (+2.4%) | 405.40 (+8.5%) | Holding Alphabet and raising the protective stop to 343.50 rather than trimming, despite a second straight bearish reading from the model. The stock rose 4.9% on Monday - a second big up-day - after Morgan Stanley reassured investors about the returns on its enormous AI infrastructure spending, and it now trades above every moving average we track with a twelve-week projection 8.5% above the current price. The bearish label came from a brief intraday poke above the 20-day high that settled 0.85% lower, on a day the shares gained nearly 5% - a mechanical trigger, not a change of trend. The fundamental picture cuts both ways and both sides are on the record: cloud revenue up 82% with a record $514bn backlog, against capital spending guided to -205bn, an 75bn raise and free cash flow at -9bn. We take the gain protection through a higher stop instead. OPERATIONAL: the 08-01 pre-committed re-arm ('a 2nd consecutive bearish print') DID FIRE - US-18/34 satisfied, US-19 calls a stable SELL-REDUCE immediately executable, adj 70, G1/G2/G3 all permit. DEMOTED on three stated grounds: (i) news OVERRIDES with a dated identifiable catalyst; (ii) the row's pillars contradict its own label (Trend Bullish 83, Consolidation_Bullish, above EMA20 347.56 / EMA50 352.38 / EMA100 346.02 / EMA200 321.19, RSI 61.1, not extended); (iii) the sell case is DECAYING not building - conviction and Sell_Rank while price rose 4.88%, the same perverse direction-of-travel the book declined on NVDA's tick (b) on 08-01. ADR-0012 armed and GENUINE (T12_Progress 92.13, Gain_ATR 16.74, Unreal_PnL_Pct +203.67% - passes US-26 cleanly) with ZERO ticks fired: tick (c) alive by a wide margin (Est_12W 405.40 vs the 380.98 threshold), tick (b) -6 not -20, no 2nd warning code, no NO-TRADE run -> armed-but-not-fired resolves to a MANDATORY trail-raise, which is this card. ADR-0020: peak 358.00 -> live 329.00 = -8.10%, below the 15% threshold, no fire. Overlay '<345' stayed HEALED (+8.3% clear) but is 30 days stale - US-31 strips originating force regardless. Est_12W 405.40 exceeds the DCF anchor 145.29 - projection assumes growth beyond priced-in. RE-ARM, tightened and pre-committed: a THIRD consecutive settled STRONG SELL - REDUCE, or any settled close below 343.50, fires the 30% trim without further debate. Note the bought off-card @328.00 on 08-01 are +13.9% at this mark. · news |
| NBIS | Nebius Group, N.V. | 212.58 | HOLD | 212.58 | 165.00 | 220.46 (+3.7%) | 250.48 (+17.8%) | 304.36 (+43.2%) | Holding Nebius and lifting the stop to 165.00 after an 11.6% jump - the portfolio's biggest move of the day - and declining the sell signal that appeared alongside it. The company secured a debt facility backed by its GPU fleet, expansion without diluting shareholders, and strong data-centre demand from Microsoft and Meta has quieted fears that AI spending is peaking; inclusion in the Nasdaq-100 is also drawing near. The stock reclaimed both its 20- and 50-day averages in a single session and our model projects meaningful upside over the coming months. Selling into that would repeat a mistake this portfolio has made before. Second-quarter results land 12 August. OPERATIONAL: STRONG SELL - REDUCE cv69 Sell_Rank 2 (adj 72) DECLINED on two independent grounds - US-18/34 lone flip (J30C/J31C were NO TRADE cv0, J29C HOLD), and US-30 (this session IS the stabilization print, reclaiming EMA20 199.53 and EMA50 205.08 together). Pillars are outright bullish (Mostly_Aligned, Trend 64, Momentum 67) while the label says sell, and Stop_Price prints 212.58 = the close exactly - the US-29 artifact verbatim. Index-inclusion catalyst -> catalyst-aware hold: never pre-trim into it. Not armed (T12_Progress 69.84). 12wk band [122.07-621.90] is 5.1x wide and near-uninformative - weight the 1wk (+3.71%). Overlay Meta-deal override UNTRIGGERED and, at 30 days stale, could not originate anything anyway (US-31). ADR-0020: original peak 270.00 (06-23) but a size cut followed (500/650 -> 300 c. 07-02) so the watermark RE-BASES to 200.00 (07-09); live 153.00 = -23.50%, which fires on the re-based watermark and is therefore SUBJECT TO THE VETO STACK - vetoed here by US-18/34 + US-30. Trail raised (~1.5 ATR below close on a 31.24 ATR, above EMA200 155.09) - repairs part of the watermark gap without inviting a whipsaw. · news |
| LLY | Eli Lilly & Co | 1121.36 | HOLD | 1121.36 | 1155.71⚠ | 1137.92 (+1.5%) | 1123.07 (+0.2%) | 1161.16 (+3.5%) | Holding Eli Lilly through Wednesday's second-quarter results rather than acting on a stop that is now marginally breached. The shares fell 2.4% to 1121.36, about 3% under our trailing stop, but they remain just above the 1120 line we committed to in advance as the level that would force a decision - and the model still projects a modestly higher price three months out. Analysts expect per share on 26bn of sales, up 30% on the year, and Goldman Sachs carries a Buy with a target. Selling a day before the print, on a breach of barely a dollar, would be trading the calendar rather than the business. OPERATIONAL: MONITOR STOP BREACH - close 1121.36 vs the 1155.71 trail = -2.97%, 3rd consecutive session and now past the 2% marginal band, so it leads the monitor and is a real breach. Not actionable on four counts: (i) the pre-committed action line is 1120.00 and the close is ABOVE it; (ii) L2 needs both legs and the 12wk is ALIVE (Est_12W 1161.16 vs the 1143.79 threshold, +3.55%); (iii) label is NO TRADE cv0 ('Consensus conflict + low conviction') and US-17 bars a bare NO TRADE from originating; (iv) PROFIT_ARMED is UNVERIFIABLE - no Cost in usportfolio.json, Gain_ATR/Unreal_PnL_Pct null, so US-26 declines the arm by rule. Q2 08-05 BMO is LLY's OWN earnings one session out -> catalyst-aware hold applies directly: trim only on a confirmed lower-high or a decisive break. Trail 1155.71 STANDS (a trail is never lowered; no model Stop_Price printed tonight). Lost EMA20 1171.10; holds EMA50 1137.47 / EMA100 1086.24 / EMA200 1020.19. ADR-0020: peak 1180.30 -> 1155.71 = -2.08%, no fire. ACTION THRESHOLD unchanged and pre-committed: a decisive settled close below 1120.00 AFTER the print. · news |
| NET | Cloudflare, Inc. | 282.74 | HOLD | 282.74 | 260.00 | 283.91 (+0.4%) | 308.19 (+9.0%) | 318.08 (+12.5%) | Holding the new Cloudflare position, bought off-card at 258.00 and now 9.6% higher, and setting its first protective stop at 260.00 - two dollars above cost, so the trade is protected into a scratch rather than a loss. The shares trade above every moving average we track with the trend reading firmly positive, and our model projects roughly 9% over four weeks and 12.5% over three months. The signal itself is only a watch-grade reading, not a fresh buy trigger, so this is a hold-and-protect rather than an add. WARNING DATA DEFECT - THE SCAN CANNOT SEE THIS POSITION: input/usportfolio.json contains TWO NET rows ({, Cost 258.0} and, 23 entries later, {}); the loader takes the LAST, so the anchor snapshot prints Held=N with Avg_Cost/Gain_ATR/T12_Progress all null and the position is invisible to every held-name code path - no Exit_Warning, no ADR-0012 arming, no stop monitoring, no ADR-0020 watermark. The monitor's unledgered_drift flag on NET is this defect surfacing, not a reconciliation problem. ARM is duplicated the same way (two -0 rows, harmless only because both are zero). MUST be de-duplicated in input/usportfolio.json before the next run; until then NET is risk-managed by hand. Signal detail: WATCH cv63 FailedBreakdown_Up_20D, 4th consecutive print of that family (WATCH/WATCH/BUY/WATCH), pillars Consolidation_Bullish Trend 69, above EMA20 267.93 / EMA50 252.03 / EMA100 235.71 / EMA200 221.62, RSI 63.5. Model Stop_Price 275.34 rejected as too tight (0.50 ATR below close); 260.00 is ~1.5 ATR on a 14.80 ATR. MoS_FV -940.26 is DEGENERATE at source (systemic finding 5) - no valuation weight taken. FOR THE LEARNINGS REVIEW: this was an off-card entry into the FailedBreakdown_Up_20D|buy cell (-7) on a name this book scored adj 51 / F=3 / MoS -940 / Quality_Pass False twenty-four hours earlier. It is working, and it was unauthorised; both facts belong in the record. |
| MU | Micron Technology, Inc. | 829.50 | HOLD | 829.50 | 770.00 | 832.07 (+0.3%) | 930.37 (+12.2%) | 1062.71 (+28.1%) | Holding Micron with the stop unchanged at 770. The shares edged up 0.8% to 829.50 and sit comfortably above that line, still 3.5% below our cost but with the model giving no actionable read in either direction for a fourth straight session. AMD's results tomorrow evening carry a direct read-through to high-bandwidth memory demand, which is the single biggest swing factor for this position. Our three-month projection is high but the range around it is so wide as to be uninformative, so we are trading the stop, not the forecast. OPERATIONAL: 4th consecutive NO TRADE cv0 - US-17 bars a bare NO TRADE from originating a dispose. Close 829.50 = +7.7% above the 770 line and above EMA100 773.84; below EMA20 899.99 / EMA50 890.13. Not armed (T12_Progress 78.06, Gain_ATR -0.32 - and US-26 would disqualify a negative-Gain_ATR arm regardless). 12wk 1062.71 = +28.1% but the band [465.74-1870.46] is 4.0x wide (systemic finding 6). ADR-0020: original peak 1102.16 (07-01) but a size cut followed c. 07-02) so the watermark RE-BASES to 947.75 (07-10); live 770.00 = -18.75%, which fires on the re-based watermark and is therefore SUBJECT TO THE VETO STACK - vetoed here by US-17 (4x NO TRADE) and US-30 (price sits +7.7% above the line, no break). Standing note: if MU ever prints an actual sell signal, that eroded watermark AMPLIFIES it. Stop cannot be raised without over-tightening - 1.5 ATR below the close on a 94.34 ATR is 688, well under the existing 770. |
| OKTA | Okta Inc. | 141.57 | HOLD | 141.57 | 135.00 | 142.41 (+0.6%) | 145.19 (+2.6%) | 161.67 (+14.2%) | Holding Okta and nudging the stop up to 135.00. It is the one position in the book whose signal strength has improved every single session for four running, it sits just above cost and above its 20-day average, with the 50-, 100- and 200-day averages stacked well below - a clean uptrend. The model projects roughly 14% over three months. Nothing here calls for action beyond tightening the protection under it. OPERATIONAL: 4th consecutive FailedBreakdown_Up_20D with conviction climbing -> - the only monotonically strengthening held name in the book. Close 141.57 (-0.25%), +1.27% over the 139.80 cost. Above EMA20 139.13; EMA50 127.51 / EMA100 113.20 / EMA200 102.59. Forward +0.59% / +2.56% / +14.20%. Not armed (T12_Progress 87.57, Gain_ATR 0.28). Model Stop_Price 138.44 rejected as too tight (0.50 ATR below close); 135.00 is ~1.05 ATR below and under EMA20. ADR-0020: peak 132.00 = the prior live stop, no erosion, no fire. Note the FailedBreakdown_Up_20D|buy cell is -7 - this is a HELD name being trailed, not a new entry, so the buy-side haircut does not apply. |
| SPCX | Space Exploration Tech Corp | — | HOLD | — | 135.00 | n/a | n/a | n/a | Holding SpaceX on the manual stop into the two biggest events of its short public life, both of which land this week. The company reports for the first time ever tonight after the close - no reporting history, no guidance track record and no consensus to anchor against, so subscriber economics, launch cadence and cash flow all get priced in one go. Two trading days later, on 6 August, roughly 911.5 million shares worth about $116bn become eligible to trade, more than doubling the free float; Elon Musk's controlling stake and key executive holdings stay locked until June 2027. Views are polarised - one side expects a heavy selloff, Morgan Stanley argues the fundamentals are largely unchanged and the pressure creates an attractive entry. Neither is tradable on a position we cannot price, so the stop does the work. OPERATIONAL: permanently MUTED - private/unquoted, absent from every scan (no_close), never synthesized, but news-assessed every run per the edge-case rule. Horizon bands: earnings +/-18%, lock-up -25%/+10%. Manual stop 135.00 = the 12-Jun IPO price. A scan Close < 135.00 AUTO-UNMUTES and becomes a full-exit trigger; a scan Close appearing at all resumes full signal+fusion treatment. · news |
Outcome & track record
Accuracy at the time of this record.
112 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.