Recommendations
0 to acquire · 1 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 |
|---|---|---|---|---|---|---|---|---|---|
| DELL | Dell Technologies Inc | 434.78 | SELL 50% | 434.78 | 406.44 | 430.98 (-0.9%) | 491.72 (+13.1%) | 576.43 (+32.6%) | We are selling half our Dell position to honour a stop-loss that has already been broken. Dell has now closed below its risk line for three straight sessions, and although the shares bounced modestly today they were sold into all day - closing 1.9% under the session's average price while our conviction score fell from. We keep the other half because the forward view is still constructive: the 12-week projection sits 33% above today's price, the shares remain above their 50- and 200-day averages, and Dell reports on September 3 into a market positioned for another earnings beat. This is a risk decision, not a verdict on the business - the valuation screen actually rates Dell the best-quality holding in the book. *** OPERATIONAL: this is a RE-FIRE, not a new decision. The identical 50% exit was armed on 08-20 for the 08-20 open and NEVER EXECUTED - portfolio JSON still reads 713, trades_detected empty, ustech_executor_state.json last_run_date 2026-08-18 = THREE sessions stale. UT-3 escalation applies: market-on-open, do NOT re-recommend as prose (L3). Measured cost of the delay: ref 437.55 -> 434.78 = worse on; realized deepens - -> -. Second measured execution-gap cost in book history (first was UT-11's under-filled TSM exit). *** ORIGINATION unchanged = decisive stop break of 455.32 (CLAUDE.md s9), close 434.78 still -4.51% below it. HALF NOT FULL: L2 INCOMPLETE - stop leg fires, forward-view leg does NOT (T12 576.43 = +32.58%); close above EMA50 414.67 / EMA200 280.31, below EMA20 448.13; RSI 48.47; Consolidation_Bullish; conviction -14 is short of the -20 collapse threshold. *** G2 SUPPRESSED the valuation amplify (MoS -3.178 <= -3 would bump : exceeds-DCF-anchor flag (T12 576.43 vs DCF/sh 485.61, +18.7%), hyper-growth, and FV_method EPV means MoS_FV measures EPV/sh 112.78 vs DCF/sh 485.61 = 4.31x intra-model spread. val_adj 0; valuation argues the OTHER way (FAIR_VALUE, Bargain True, Quality_Pass True, F 8, Quality_Rank 7). *** ADR-0020 does NOT fire but is CONVERGING: the 08-20 peak_stop re-base to 406.44 was contingent on a cut that never happened, so the honest watermark is 455.32 vs live 406.44 = 10.73% trail erosion with ZERO size reduction - 4.3 pts from the mandatory 15% trigger. Executing today resolves it and legitimately re-bases peak_stop to 406.44. *** RESIDUAL STOP HELD AT 406.44 (ratchet: today's 2.0 x ATR below the 443.73 high computes to 386.79, lower, so 406.44 stands); 6.97% / 1.00 x ATR below close, just under EMA50; residual risk 356 x 28.34 = = 1.05% of. Engine Stop_Price 420.54 is again only 0.50 x ATR below close - systemic finding -20 reproduces. *** Proceeds; residual = = 16.01% of (from 32.15%). 12wk EXCEEDS DCF ANCHOR. f = 0.134 (18 graded). UT-11: re-fire any quantity shortfall next session. · news |
| AAOI | Applied Optoelectronics, Inc. | 129.10 | HOLD | 129.10 | 111.93 | n/a | 162.34 (+25.7%) | 120.41 (-6.7%) | Holding Applied Optoelectronics after a strong reversal session: the shares fell to 118.60 intraday then closed at 129.10, up 5.7% and near the day's high on heavy volume, cutting the position's loss from 18.6% to 14.0%. The optical-transceiver story behind it is intact - management guides to roughly 1bn of revenue this year with 800G and 1.6T demand exceeding capacity into mid-2027, and one analyst carries a target - though the shares remain speculative and expensive on current earnings. Our stop stays at 111.93, now comfortably 15% below the price, and the position is deliberately half-sized because of that volatility. *** L2 STAYS INCOMPLETE: forward-view leg still armed (T12 120.41 is -6.73% BELOW the close 129.10) but the stop leg is now FAR away (close +15.34% above 111.93, up from +9.17%) - no exit, and the leash got materially longer. A decisive close below 111.93 still completes L2 and UT-9 would not block it. *** UT-9 CONTROLS THE ARM, NINTH SESSION: PROFIT_ARMED prints while Gain_ATR is -1.26 and the position is -13.95% - arm-on-loss mislabel, INERT. *** Signal re-formed to FailedBreakdown_Up_20D cv49; engine stop 95.86 is far below the manual 111.93, which HOLDS. *** Valuation can neither originate nor amplify: Val_Signal NA, FV_per_share -3.38 (negative EPV - the screen cannot value the company), Quality_Pass False, F 4. *** Position = 4.58% of; open risk to stop (129.10 - 111.93) x 342 =; total risk from cost = 1.35% of. ADR-0020 peak 111.93 = live, erosion 0.00%. News two-sided but no override (SA buy vs a Yahoo 93%-overvalued piece); both sides were known at entry. · news |
| NVDA | NVIDIA Corporation | 216.85 | HOLD | 216.85 | 206.05 | n/a | 227.52 (+4.9%) | 236.63 (+9.1%) | Holding Nvidia into next Wednesday's results rather than adding or trimming. The shares were quiet at 216.85 while the rest of the AI hardware complex swung sharply, and they sit above their 20-, 50- and 200-day averages with a 12-week projection about 9% higher. The setup into the print is favourable - the company guides to roughly $91bn of quarterly revenue at 75% gross margin, Jefferies expects the largest revenue beat in its history, and the stock trades on a low-20s forward earnings multiple, unusually cheap for this point in the year. The reason we are not adding is risk, not conviction: our stop at 206.05 now sits inside the roughly plus-or-minus 6.75% move the options market is pricing for the result, so an ordinary adverse reaction would close the position. *** UNCHANGED WATCH from 08-20, stated verbatim: implied band off 216.85 is 202.22 - 231.57 and the 206.05 stop is INSIDE it. Stops ratchet up only so 206.05 HOLDS (engine Stop_Price 206.18 is the same level); REVISIT AT THE 08-25 CLOSE, the last session before the print. *** UT-9 CONTROLS: PROFIT_ARMED at Gain_ATR -1.86 on a -4.39% position - arm-on-loss, INERT, never a profit-take. L2 fails BOTH legs (close +5.24% above stop; 12wk positive). Conviction is -7, not a collapse. *** L6 STANDS: no add before 08-26; the horizon's add-after-print read-through to MU/KLAC/NBIS/NET is a POST-print licence, not a pre-print one. Position = 3.78% of; open risk (216.85 - 206.05) x 168 =. ADR-0020 peak 206.05 = live, erosion 0.00%. · news |
| MU | Micron Technology, Inc. | 974.33 | WATCH | 974.33 | 929.07 | n/a | 1091.41 (+12.0%) | 1080.36 (+10.9%) | Micron was the closest we came to a purchase and we passed by a narrow margin. The shares rose 4% to 974.33 on the day, rank first on our buy list, and Micron screens as the single highest-quality business in this book - but the price already embeds a great deal of the memory upcycle, and that valuation haircut leaves the trade three points short of our threshold. There is also a better entry point coming: Nvidia reports next Wednesday and its commentary on memory demand is the cleanest read-through to Micron's high-bandwidth memory order book, so the disciplined route in is after that print, not three sessions before it. The news is genuinely two-sided - a bull target and 16 take-or-pay contracts through 2030 covering roughly $100bn against Citi cutting to on a 2027 price peak and SK Hynix adding $38bn of competing capacity. *** ARITHMETIC: raw cv 69 (FIRST raw print above the 65 floor since 08-19), val_adj = clamp(round(-1.349 x 5)) = -7 -> adj 62. Entry_Distance_ATR -0.0 so this would be a MARKET buy, not a queue - no chase problem. Quality_Pass True, F 8, Quality_Rank 1 - the gate WOULD pass. Closed +10.38% above VWAP near the session high (977.28). *** TWO INDEPENDENT REASONS NOT TO FORCE IT: (1) L1/UT-12 - FailedBreakdown_Up_20D|buy is on the discount list (own-book n=2, hit_1w 0.50, hit_4w 0.00) and is now the modal signal on this book; (2) L6 + horizon add-after-print. *** RE-ARM (written on PRICE as well as conviction, per systemic finding -20): raw conviction >= 72 at the current MoS, OR a pullback lifting MoS_FV above -1.0. *** SYSTEMIC: MU's FV_method is EPV (EPV/sh 432.15 vs DCF/sh 772.94 = 1.79x intra-model spread) - the same EPV/DCF divergence G2 suppresses on the sell side is applied at FULL weight here on the buy side. See systemic finding 8. · news |
| NBIS | Nebius Group, N.V. | 220.11 | WATCH | 220.11 | 207.62 | n/a | 270.06 (+22.7%) | 268.08 (+21.8%) | Nebius returned to our buy list for the first time in a week but does not clear the bar. The neocloud shares slipped 1.7% to 220.11 while the rest of the AI hardware complex bounced, and the valuation is among the most stretched in the book - the shares trade at many multiples of any defensible intrinsic estimate. They also fail our quality screen outright, which blocks new positions regardless of the signal. We continue to watch rather than buy. *** ARITHMETIC: raw cv 66 (above the floor), val_adj -10 (MoS -11.39) -> adj 56; separately Quality_Pass False (F 6) BLOCKS new entries outright. Two independent failures. *** The stand-down that began 08-15 at cv88 has now saved roughly 21% cumulative (277.68 -> 220.11). Overlay override ('invalidated when the Meta deal is materially at risk') is moot. Horizon carries an add-after-print read-through from NVDA 08-26 - a post-print licence only (L6). |
| PLTR | Palantir Technologies Inc. | 173.96 | WATCH | 154.81 | 148.31 | n/a | 164.47 (+6.2%) | 247.89 (+60.1%) | Palantir screens as a buy on our preferred pullback signal but the price has run far beyond the level that signal wants us to pay. The shares closed at 173.96, roughly 12% above the model's entry point of 154.81 and 12.4% above the session's average price - buying here means paying up for a setup that was designed to be bought on weakness. The business quality is genuinely good; the valuation is the second-most stretched in the book. We wait for the pullback rather than chase. *** ARITHMETIC: raw cv 64 (below the 65 floor pre-haircut), val_adj -10 (MoS -13.011) -> adj 54. Entry_Distance_ATR -2.95 is the most extreme extension on the board - UT-6 no-chase at its clearest. Quality_Pass True, F 8, Quality_Rank 5 - quality is fine, price is not. Note the signal family is PullbackBuy_VWAP (L1's FAVOURED cell) and it STILL fails, which is the no-chase rule correctly overriding a favoured family (UT-6). |
| SNOW | Snowflake Inc. | 321.29 | WATCH | 308.15 | 297.13 | n/a | 324.53 (+5.3%) | 457.84 (+48.6%) | Snowflake fails on the same two counts as last session and we pass again. Conviction of 64 is already below our entry threshold before any valuation adjustment, the shares are priced well above our intrinsic estimate, and the company does not pass our financial-quality screen - which blocks new positions outright. Results are due around September 2, which is another reason not to open a position on a marginal signal now. *** ARITHMETIC: raw cv 64 is BELOW the 65 floor PRE-haircut; val_adj -10 (MoS -1.946) -> adj 54; Quality_Pass False (F 5, Quality_Rank 20.5) blocks new entries. Entry_Distance_ATR -1.19 means a queue not a market buy even if it qualified. RE-ARM: conviction >= 75 AND the quality screen clearing. |
| MSFT | Microsoft Corp | 481.15 | WATCH | 464.17 | 454.26 | n/a | 478.90 (+3.2%) | 631.65 (+36.1%) | Microsoft re-entered our buy list but at a lower conviction than when it left, and it still fails the financial-quality screen that blocks new positions. The shares closed at 481.15, about 3.7% above the model's entry level, so this would be a wait-for-pullback order in any case. We pass. *** ARITHMETIC: raw cv 62 is BELOW the 65 floor PRE-haircut (and LOWER than the cv76 at which the gate blocked it on 08-19 - it re-entered the pool weaker); val_adj -8 (MoS -1.653) -> adj 54; Quality_Pass False (F 6, Quality_Rank 16) blocks new entries. Entry_Distance_ATR -1.72. This closes the d-iii measurement loop further: the 08-19 headline gate block has now decayed twice and cost the book nothing. |
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 →
0.134 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.