US Semiconductors — End-of-Day Verdict
1 ACQUIRE, 0 DISPOSE | NVDA BUY day-limit <=220.00 (entry 218.99) pending the 08-07 open - the book's FIRST since inception, at 0.75x model size and 0.096% of book risk, R:R 4.0:1 | AMD HOLD 172, escalation tested and NOT fired (489.28 > 475.00, +3.01%), stop 475.00 re-armed unchanged | TSM HOLD, TRAIL RAISED 382.61 -> 385.57, catalyst-aware hold into the ~08-10 TSMC July revenue print
Recommendations
1 to acquire · 0 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 |
|---|---|---|---|---|---|---|---|---|---|
| NVDA | NVIDIA Corporation | 218.99 | BUY | 218.99 | 214.97 | 223.81 (+2.2%) | 235.07 (+7.3%) | 226.27 (+3.3%) | We are buying NVIDIA - the first capital this book has put to work since it was funded, and it is going into the only bullish setup on a twenty-name board. Yesterday the same stock printed an early 'anticipation' signal and we passed, because it had already run 3.4% on a headline and we do not chase. Today it did the thing that answers that objection: it converted to a full buy signal and consolidated rather than extended, so the model's entry price and the market price are now the same number,. The support is real - analysts covering the stock rate it a buy, none rate it a sell, and the average price target sits near with its next quarterly report on 26 August expected to show roughly 96% growth. We are sizing this deliberately small, at three-quarters of the model's recommendation, for two honest reasons: this particular signal family has never worked in this book (three attempts, three stopped out), and NVIDIA now backstops around $250bn of debt tied to OpenAI's data centres, which means it is helping finance its own customers. The trade earns its place on geometry rather than confidence - the stop sits just 1.8% below entry while the target is 7.3% above, so a fourth consecutive failure costs the book under a tenth of one percent, and one success pays four times the risk. OPERATIONAL: signal FailedBreakdown_Up_20D cv68 (from BUY - ANTICIPATE cv83 08-05), Buy_Rank 1, Entry_Distance_ATR -0.00 (at the line, no-chase bar met on the tape). USC-3 gate RUN, NOT OVERRIDDEN: the cell is 3/3 stop-first (hit_1w 0.0, median_err_1w -9.77%) and report-only at n=3, read as stop-prone not neutral; the escape clause is a PullbackBuy_VWAP confirmation OR a passing quality gate, and Quality_Pass is True - the only mandate name that both prints a buy label and clears the F-Score/FCF screen - so full process is permitted and no gate override was taken. val_adj SUPPRESSED to 0 under G2/note B (DCF anchor vs a price is a broken hyper-growth artifact and must not originate the -10 floor haircut); the size modulator is deliberately LEFT LIVE at 0.75x (MoS_FV -2.16, the -3<MoS<=-1 tier) - note B is written narrowly against a conviction VETO, not against valuation informing size, and on a 3/3-stop-first family the conservative reading is the right-signed error. Sizing: 35.06% x x 20% buy_pct = -> x0.75 = -> @ 218.99 = (5.2% of book); max_position_pct 60% and both non-binding. Risk 235 x = = 0.096% of book; reward; R:R 4.0:1. Entry band 217.98-220.00 (218.99 +/- 0.25xATR 8.04); VEHICLE IS A DAY-LIMIT AT 220.00, not a market order, so an 08-07 gap-up mechanically cannot fill (L8 repricing + USC-4 no-chase enforced by order type). Estimates: 1wk 223.81 at f=0.30 MANDATORY (L5, calib_1w -0.548 degenerate); 4wk 235.07 = Target_Price, the risk objective, and it CARRIES THROUGH the 08-26 Q2 FY27 print (upside optionality and event risk in the same window); 12wk 226.27 WARNING exceeds DCF anchor - projection assumes growth beyond priced-in - and prints BELOW the 4wk, an inverted curve logged as systemic finding #3, so treat it as the weakest of the three. Honest counts against: conviction fell on the conversion leaving only 3 points over the 65 floor; 92% of revenue is data-centre concentrated in a few hyperscaler capex budgets. AUTO-WITHDRAW: do NOT reprice the limit upward; withdraw outright if the 08-07 settled close prints conviction < 65, the label reverts to WATCH/NO TRADE/AVOID, or expires_on 2026-08-11 passes unfilled. Horizon: Nonfarm Payrolls 08-07 is the fill session - the one real execution risk. · news |
| AMD | Advanced Micro Devices, Inc | 489.28 | HOLD | — | 475.00 | n/a | n/a | 501.95 | We are holding AMD for a second session, and again the reason was written down before today rather than argued after it. The line we committed to was that a decisive close below triggers a full exit, judged on the settled close. AMD closed at - 3.0% clear of that line, and up 1.5% on the day with no new company news, which is the mildly encouraging read after last week's earnings selloff. The quarter behind that selloff was a record one that the market simply refused to reward: revenue up 50%, data-centre sales more than doubled, guidance raised above consensus, and Wells Fargo, Jefferies and JPMorgan all lifted their price targets afterwards. Nothing in the last 24 hours changed the case in either direction, so the position and the line both stand unchanged. OPERATIONAL: origination test run on every branch and failed on all. Signal-based (sec9): anchor prints AVOID on the bearish FailedBreakout_Down_20D family - not a sell label - and the book's only actionable scorecard cell (NoTrade|sell, bias -7, hit_1w 0.167, deteriorated from -5/0.25) instructs directly against originating sells from non-sell labels here. L2: fails BOTH legs - 489.28 is +3.01% / +0.36xATR above the 475.00 stop, and the 12wk 501.95 clears the 1.02x dead-view threshold 499.07 by, alive but the THINNEST margin yet recorded - a 12wk slipping under 499.07 while price holds above the stop would fire ADR-0012 tick (c) on the next print, so watch it. ADR-0012: PROFIT_ARMED is a FALSE arm - position is -5.19% below cost with Gain_ATR -0.68 and T12_Progress 97.48; no profit exists to take, twelfth straight session of this defect (learnings advisory (a)). ADR-0020: Peak Stop 526.88 vs live 475.00 = -9.85% against the 15% threshold, no size cut since the peak; MANDATORY VETO-IMMUNE 30% TRIM TRIGGERS AT A STOP BELOW 447.85 - recorded for the fourth consecutive session so it cannot arrive unannounced. News: nothing new on 08-06; the 08-05 negatives (SpaceX/Vera Rubin exclusivity locking out MI450, capex vs, soft H2 PC) cannot originate under sec9 and there is no proposal to demote. Trail check: 2.0xATR chandelier off 489.28 = 410.13, sixty-five points BELOW the standing line, so 475.00 is RETAINED, not loosened. Escalation re-armed unchanged for the 08-07 settled close. · news |
| TSM | Taiwan Semiconductor Manufacturing C | 418.20 | HOLD | — | 385.57 | n/a | n/a | 386.72 | We are keeping the small TSMC core and tightening the safety net underneath it rather than selling into a catalyst. The stock rose 1.0% today, and because our trailing stop is calculated from the price, that rise lifts our exit line from to - the first time this position's stop has ratcheted upward since we opened it. Stops only ever move one way. The reason we are still holding at all is two sessions away: TSMC reports July monthly revenue around 10 August, and it is the single highest-frequency read the market gets on whether AI capital spending is still accelerating. June came in 68% above a year earlier and the company has raised full-year growth guidance above 40%. We already took the risk down hard - 78% of this position was sold last week at - so what remains is a deliberately defined core sized to carry exactly this event, not a leftover we forgot to sell. OPERATIONAL: trail raise 382.61 -> 385.57 = 2.0xATR chandelier off the 418.20 close (418.20 - 2 x 16.313), above the standing line for the first time in this position's life; new stop sits 7.8% under price, a materially tighter leash into the event. ADR-0012 tick (c) DOES fire - 12wk 386.72 is far under the 426.56 dead-view threshold - but the arm beneath it is the 24TH CONSECUTIVE FALSE ARM (position -4.52% below cost, Gain_ATR -1.21, T12_Progress 108.14); a profit-take on a position with no profit is not a mechanism this framework has, and armed-but-not-fired resolves to the MANDATORY TRAIL-RAISE, which is what is being done. L2 fails leg 1 (418.20 is +9.30% above the stop). Signal is AVOID cv57 on the bearish FailedBreakout_Down_20D family - not a sell label - and the NoTrade|sell cell at bias -7 argues against originating a sell from it. val_adj suppressed to 0 under G2 (DCF anchor against a price is broken). CATALYST-AWARE HOLD BINDING: TSMC July monthly revenue ~08-10, two sessions; per USC-5 a DEFINED core - not a discretionary majority - rides the catalyst, and the clause permits a trim only on a confirmed lower high or a decisive stop break, neither of which exists. Position value. · news |
Outcome & track record
Accuracy at the time of this record.
18 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 →
-0.548 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.
53 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.