Recommendations
0 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 |
|---|---|---|---|---|---|---|---|---|---|
| DELL | Dell Technologies Inc | 468.65 | HOLD | 468.65 | 455.32 | n/a | 528.15 (+12.7%) | 667.83 (+42.5%) | Dell fell 2.3% in a session where the entire AI-hardware complex was sold, and it was one of the better performers in that group - our optics and memory names fell three to seven times as hard. Nothing changed at the company: full-year revenue guidance stands at $167bn, AI-server revenue is running at 1bn and the AI order backlog is 3bn, and the stock is still up sharply on the year and within 9% of its record high. It remains the only holding our own valuation work flags as a genuine bargain on fundamentals. We hold the full position, keep the protective stop where it is, and note that quarterly results on 3 September are now the next real test. OPERATIONAL: no sell origination class fires. Conviction flat; Exit_Warning blank; NOT profit-armed (Gain_ATR 0.25 < 3.0, T12_Progress 70.18% < 90) so ADR-0012 is unavailable; L2 fails both legs (close +2.84% above stop, 12wk 667.83 = +42.5%); G3 forward-view gate blocks any trim. Valuation is amplify-only and here argues a trim DOWN a tier (FAIR_VALUE, Bargain True, F-Score 8) - there is no trim to scale. STOP HELD at 455.32, NOT raised: chandelier 2.0 x ATR (29.75) below the 514.00 ATH = 454.50, which is BELOW the existing ratchet, and the engine's own Stop_Price fell to 453.77 - stops ratchet up only, so both are rejected. ADR-0020: peak_stop 455.32 = live stop, trail_erosion 0.00%. WATCH - the close is now only 0.45 x ATR above the stop on a 6.35%-ATR name, the tightest this line has ever sat; one ordinary down session takes it out, and if that break comes with the 12wk still at +42.5% it is L2-incomplete, so judge it on decisiveness, not on the touch. Dell FQ2 ~2026-09-03 is now 10 sessions out. · news |
| AAOI | Applied Optoelectronics, Inc. | 131.41 | HOLD | 131.41 | 111.93 | n/a | 167.39 (+27.4%) | 135.97 (+3.5%) | Our optics position fell 15% in a single session and is now 12.4% below cost, giving back the day-one gain and more. There was no news about the company. The whole AI-hardware trade was sold after a Wall Street Journal report that nine large technology firms carry around trillion of off-balance-sheet AI commitments, on the same day bond yields rose and two leading AI labs reported softer revenue than expected - memory, optics and chip-equipment names fell together. This is a very high-beta stock, so it moves several times the sector, and that was known and is why the position was taken at half size with a wide stop. The stop has not been threatened and the position is small, so we hold. What we are watching is not the price but the forward view: our twelve-week projection has fallen to barely above the current price, and if it turns negative the case for holding goes with it. OPERATIONAL: no sell origination class fires. Row prints HOLD cv49 (from 51, -2, far from the -20 collapse tick). L2 fails - close 131.41 is +17.4% above the 111.93 stop; 12wk 135.97 = +3.47%. *** UT-9 CONTROLS: the row prints Exit_Warning=PROFIT_ARMED while Gain_ATR is -1.04 and the position is -12.41% - an arm-on-loss mislabel, INERT, never read as a profit-take. Worse than the TSM precedent: T12_Progress jumped 26.1% -> 96.65% NOT because the position gained but because the 12wk target COLLAPSED 168.66 -> 135.97 (-19.4%) toward the falling price. The arm metric rises as the thesis deteriorates. *** Dead-view tick (c) MISSES BY : threshold 1.02 x close = 134.04, Target_12Week = 135.97. Even if it fired, ticks require an ARMED position and UT-9 rules this arm inert - so the tick is moot, but flag it. STOP HELD 111.93; the engine's chandelier collapsed to 95.43 as ATR blew out 16.57 -> 17.99 - rejected, stops ratchet up only. ADR-0020: peak 111.93 = live, erosion 0.00%. Risk profile changed materially without a trade: the stop was -27.7% below the close at entry and is now only -14.82% (1.08 x ATR) - the position tightened itself. Open risk from here (131.41-111.93) x 342 =; from cost (150.03-111.93) x 342 = = 1.32% of. WATCH: one more down session flips the 12wk view negative, at which point L2's second leg arms and a stop-break becomes a genuine exit basis. · news |
| NVDA | NVIDIA Corporation | 219.74 | HOLD | 219.74 | 206.05 | n/a | 219.12 (-0.3%) | 246.90 (+12.4%) | We hold Nvidia into its 26 August results and decline the model's invitation to buy more. The signal engine has come back to life - it now rates the stock a buy and wants to add - but it wants to add at, roughly 4% below where the shares closed, and adding to a position that is already slightly underwater in the middle of a sector-wide sell-off is precisely the mistake this book has made before and written down as a rule. The set-up into the print is still the reasonable one: about $92bn of expected revenue, earnings per share roughly double a year ago, and a valuation near 25 times forward earnings against 40%-plus expected growth. The options market implies a 6.75% move on the result. Our stop sits below that implied range, which is deliberate. OPERATIONAL: row flips NO TRADE cv0 -> BUY - ADD cv71 (PullbackBuy_VWAP, Buy_Rank 2), Portion 63.29%, buy-limit @210.94, EDA -1.60. ADD DECLINED TWICE OVER: (1) arithmetic - val_adj -10 (MoS -2.56, capped) takes adj conviction to 61, BELOW the 65 floor; (2) UT-10 - do not pyramid a PullbackBuy ADD into a held name that is already underwater during a sector-wide high-ATR reset; the two DELL ADDs graded under this exact pattern both hit their stops. L6 also cautions against sizing up 5 sessions ahead of a print. No sell either: NOT armed (Gain_ATR -1.28, T12_Progress 89.0 < 90 - and UT-9 would rule an arm on a losing position inert anyway); L2 fails (close +6.6% above stop, 12wk 246.90 = +12.4%). STOP HELD at 206.05: the options market implies a 209-239 range on the print, so 206.05 deliberately sits BELOW the implied downside edge; a mechanical 2.0 x ATR chandelier off the 236.54 high would give 225.54, above the close, which would be a guaranteed shakeout. 206.05 = 2.49 x ATR below close. ADR-0020: peak 206.05 = live, erosion 0.00%. The 08-26 print decides this line; the horizon stance is hold-through with an add-after-print read-through to the complex. · news |
| MSFT | Microsoft Corp | 481.63 | WATCH | 481.63 | 475.89 | n/a | 504.58 (+4.8%) | 648.01 (+34.5%) | Microsoft was the strongest buy candidate on our board and the first name in weeks to clear our conviction bar even after we mark it down for being expensive - and we still did not buy it. It fails our fundamental quality screen, which blocks new positions outright. The tape supports the idea: money rotated out of AI hardware and into large software on the day, and Microsoft held its ground while our own hardware holdings were sold hard, with Azure now past $100bn of annual revenue and 30 million paid Copilot seats. Against that, Beijing is reportedly accelerating the removal of Windows from state-linked systems. We would rather record the miss and let the screen be measured than override it on a strong day. OPERATIONAL: cv 76 (from 64), Buy_Rank 1, at the model entry (EDA -0.00, Market order, close = Entry_Price 481.63), pillars Mostly_Aligned / Trending / Trend 72 Momentum 60. val_adj -8 (MoS -1.65) -> adj 68, CLEARS the 65 floor - the first ACQUIRE-slot name to do so since 08-13. BLOCKED SOLELY BY THE QUALITY GATE: Quality_Pass False, F-Score 6, Quality_Rank 16. Family caution applies too (L1/UT-12 discount FailedBreakdown_Up_20D chases; own-book cell n=2, hit_4w 0.0). Ledgered direction:none so the gate's cost is priced - this is the cleanest test case yet for learnings backlog item d-iii, because for once the gate is the ONLY thing standing between the book and a trade. · news |
| SNOW | Snowflake Inc. | 325.33 | WATCH | 304.59 | 292.27 | n/a | 322.91 (+6.0%) | 463.60 (+52.2%) | Snowflake screens as a buy on the pullback pattern we generally trust most, and it is the sort of software name money rotated into on the day. We pass: it is expensive against any conservative estimate of intrinsic worth, it fails our fundamental quality screen, and the model wants to buy it about 6% below the closing price rather than here. A name we cannot own on fundamentals is not one we chase up to. OPERATIONAL: cv 71, Buy_Rank 3, val_adj -10 (MoS -1.95) -> adj 61, 4 short of the 65 floor; Quality_Pass False, F-Score 5, Quality_Rank 20.5 (worst-but-one in the buy pool). EDA -1.69 = close is 1.69 ATR past the 304.59 model entry, so a queue not a market buy even if it qualified (UT-6). Volatility pillar Bearish. Results ~2026-09-02, just outside the horizon window. Ledgered direction:none for gate/floor cost measurement. |
| CRM | SalesForce Inc | 196.14 | WATCH | 196.14 | 192.15 | n/a | 212.09 (+8.1%) | 235.40 (+20.0%) | Salesforce was the best-performing name on our board, up 2.7% on the day money left AI hardware for software, and it is one of only two names we track that passes our fundamental quality screen and carries a bargain flag. It still falls short of the conviction bar once we mark it down for price, and this is the second time in two weeks we have looked at it. Worth keeping close: a modest further improvement in the signal makes it the front-runner, and unlike most of this book it would clear the quality screen. OPERATIONAL: cv 68, Buy_Rank 4, at the model entry (EDA 0.00, Market, close = Entry_Price 196.14). val_adj -10 (MoS -2.23, uncapped -11.15) -> adj 58, 7 short of the floor - valuation is what kills it, not the signal. Quality_Pass TRUE, F-Score 7, Bargain True, Val_Signal FAIR_VALUE (note the split: EPV/share 59.55 drives MoS -2.23 while DCF/share is 379.39 - the composite reads FAIR_VALUE but the MoS-driven val_adj takes the maximum haircut; flagged as a systemic finding). Family caution L1/UT-12 (FailedBreakdown_Up_20D). PRECEDENT: the only graded CRM buy in this book (2026-08-04 @185.95, same signal family) HIT its 1wk target, +6.22%, err +3.07% - n=1, report-only, not a bias. Re-arm: conviction >=75 keeps it above the floor at the current valuation haircut. |
| PLTR | Palantir Technologies Inc. | 171.54 | WATCH | 151.37 | 144.37 | n/a | 161.78 (+6.9%) | 244.44 (+61.5%) | Palantir passes our fundamental quality screen but is the second most expensive name we track relative to any defensible estimate of intrinsic value, and the shares sit 13% above their average traded price - the most stretched reading on the board. The model itself wants to buy 12% lower. There is nothing to do here at this price. OPERATIONAL: cv 64, Buy_Rank 5, val_adj -10 (MoS -13.01) -> adj 54, 11 short of the floor. Quality_Pass True, F-Score 8, Quality_Rank 5 - quality is not the problem, price is. VWAP_Distance +13.33%, EDA -2.89 = 2.89 ATR past the 151.37 model entry, RSI 65.7; textbook no-chase (L1/UT-6). Ledgered direction:none for measurement. |
Outcome & track record
Accuracy at the time of this record.
17 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.