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 |
|---|---|---|---|---|---|---|---|---|---|
| PLTR | Palantir Technologies Inc. | 185.93 | BUY — QUEUED @163.94 | 163.94 | 155.72 | 165.11 (+0.7%) | 176.16 (+7.5%) | 264.95 (+61.6%) | Palantir is compounding at a rate few software companies ever reach - Q2 revenue up 93% year on year to 94bn, US commercial up 149%, and the Pentagon has now made its Maven Smart System a formal program of record worth up to 30bn over five years, which converts a headline contract into a contracted revenue base. The engine agrees emphatically: maximum conviction 100, rank #1, on the pullback-to-VWAP setup that is historically the most reliable buy pattern. The catch is price - the stock closed 11.8% above the level the model wants to pay and trades far beyond any intrinsic anchor, so we bid at the VWAP rather than chase, and we buy only if the market comes to us. OPERATIONAL: L1 favoured family (PullbackBuy_VWAP). No-chase enforced - Entry_Distance_ATR -2.68, close 185.93 near the 186.85 session high, so GTC limit at the printed Entry_Price 163.94 (= anchored VWAP, just below EMA20 167.16), never a market order. Entry range 161.89-165.99 (0.25xATR 8.207). Quality gate PASS (F 8, Quality_Rank 6/31). val_adj -10 (MoS -13.08), size mod x0.5. Sizing: 66.95% x x 20% = x 0.50 = / 163.94 = = (6.69% of). Stability 3-of-5 snapshots in the buy pool. Withdrawn 08-25 at a 156.52 limit and the name has since run 18.8% - discipline was right on the day, re-arm watched and fired today. 12wk 264.95 exceeds DCF anchor 46.62. Thiel insider selling noted as the counter. expires_on 2026-09-02 (anchor + 3 trading sessions); auto-withdraw if conviction < 65. · news |
| CRM | SalesForce Inc | 252.05 | BUY — QUEUED @213.78 | 213.78 | 174.37 | 221.35 (+3.5%) | 292.59 (+36.9%) | 343.02 (+60.5%) | Salesforce delivered the quarter the bears said it could not - adjusted EPS of against expected, full-year profit guidance lifted 18.5%, and its AI business (Agentforce plus Data Cloud) now runs at close to 9bn of recurring revenue, up more than 210%, extended by a widened Anthropic partnership. It is also one of only two names in this book the valuation screen still calls fairly priced. But the stock rose 22.6% in a single session to an RSI of 80, and this book's own hard-won lesson is that blowout gaps round-trip. So we want the business at a sensible price, not the euphoria: a resting bid 15% lower that fills only if the gap retraces. OPERATIONAL: L6 governs - post-earnings gap +22.58%, RSI 80.0, Entry_Distance_ATR -2.47, so QUEUE at the printed Entry_Price, never chase. ORDER-TYPE ARTIFACT: row prints 'Buy Stop @ 213.78' against a 252.05 close - a stop below the market is an instant market fill, so it is queued as a GTC LIMIT at the same level (systemic finding 2). Entry range 209.90-217.66 (0.25xATR 15.505). Quality gate PASS (F 7, Quality_Rank 15/31); FAIR_VALUE, one of only two in the book. val_adj -10 (MoS -2.51 x5 = -12.55 clamped), size mod x0.75 (-3 < MoS <= -1). Sizing: 66.95% x x 20% = x 0.75 = / 213.78 = = (10.03% of, largest line after DELL). VALUATION STALE-GENEROUS: fairvalue CSV priced off the 08-24 close ~206, before the gap. 12wk 343.02 is BELOW the 379.39 DCF anchor - no annotation needed, the only acquire today that does not need one. Counter-view on record (Seeking Alpha 'dead cat bounce'). expires_on 2026-09-02; auto-withdraw if conviction < 65. · news |
| TSM | Taiwan Semiconductor Manufacturing C | 427.30 | BUY | 427.30 | 420.60 | 429.35 (+0.5%) | 448.69 (+5.0%) | 414.21 (-3.1%) | Nvidia's blockbuster quarter is, mechanically, TSMC's order book - it makes the chips. TSMC's own numbers already confirm it: July revenue up 44.7% year on year, full-year growth guidance raised to above 40%, and 67.7% gross margins that almost no manufacturer anywhere sustains. It is the second-highest-quality business in this book on the fundamental screen, and unusually it closed exactly at the price the model wants to pay, so no chasing is required. This is deliberately a four-week swing rather than a long-term hold - the twelve-week projection is slightly negative, so we take a tightly-stopped trade risking 1.6% to make 5%, and we size it small. OPERATIONAL: only non-extended entry in the pool (Entry_Distance_ATR 0.00, Order_Type Market, entry = settled close). Quality gate PASS with F-Score 9 / Quality_Rank 4. - second best in the book. RSI 55.2, above EMA20 418.30 / EMA50 417.28 / EMA200 374.67, Mostly_Aligned, Trend 76 / Momentum 67. THREE RESERVATIONS ON THE RECORD: (1) Target_12Week 414.21 is -3.06% BELOW the action price, under the framework's own 1.02x dead-forward-view threshold - hence swing-trade framing and 0.5x sizing, not a core position; (2) FailedBreakdown_Up_20D is the L1/UT-12 discounted family and TSM's own 07-07 trade in this exact family was stop-first at -6.95%; (3) single-session reversal AVOID 57 -> STRONG BUY 86 (admitted as a FRESH signal under the ACT-NOW bar, not a same-day flip-flop, but zero-stability reads have been kept out before). TAKEN BECAUSE THE RISK IS TINY AND DEFINED: stop 420.60 = -1.57%, 4wk objective +5.01%, R:R 3.2:1, dollar risk 154 x = = 0.10% of. Entry range 424.63-429.97 (0.25xATR 10.697). val_adj -10 (MoS -13.76), size mod x0.5. Sizing: 66.95% x x 20% = x 0.50 = / 427.30 = = (6.67% of). INVALIDATION STATED IN ADVANCE: decisive close below 420.60 closes it, no re-vote, no averaging down; if the 4wk objective is hit while Target_12Week is still below the close, close at the objective rather than hold. · news |
| AAOI | Applied Optoelectronics, Inc. | 113.28 | SELL ALL | 113.28 | 111.93 | n/a | n/a | 77.51 (-31.6%) | On the single best day the AI-infrastructure complex has had all month - nineteen of thirty names up, median +1.70% - Applied Optoelectronics fell 0.42% on rising volume. A business that cannot rally with its own sector on its sector's best day is not recovering, it is being distributed. The model's twelve-week view has fallen for eight straight sessions and now sits 31.6% below the current price, the shares remain below every meaningful trend line, and conviction in the setup is collapsing. The growth story is genuinely exciting - record revenue up 86%, capacity sold out into 2027 - but the company loses money, funds itself by selling new shares into exactly this kind of bounce, and ranks 28th of 31 on quality. We are down 24.5%; we take the loss and redeploy. OPERATIONAL: SIXTH re-fire of the 08-25 MOO exit (executor 8 sessions stale, last_run_date 2026-08-18); UT-3/UT-11 say re-fire, do not re-litigate. Execution gap COST this session (113.28 vs the 113.76 ref) - sixth entry in the tally (-988.89/+2606.10/-3173.73/+6536.67/+208.62/-164.16). L2 origination. (a) DEAD 12WK eighth session, T12 77.51 = -31.58%, monotonic 135.97->116.18->120.41->112.17->84.20->83.05->80.24->77.51; (b) CONVICTION COLLAPSE AVOID (-12) on a bearish family with Exit_Strategy 'immediate'; (c) STRUCTURE WIDENING - below EMA20 122.16 / EMA50 127.14, only EMA200 106.56 beneath; VWAP_Distance -12.31%, wider than -11.0%; Consolidation_Bearish, Trend 50 / Momentum 33; (d) STOP RECLAIM FADING - close 113.28 vs stop 111.93 = +1.21% = 0.11 ATR, NARROWING from +1.63%. WHY 100%: x 113.28 = = 3.92% of; a half leaves a -24.5% loser with a -31.6% forward view and an 'immediate' exit strategy; realized -vs cost 150.03. VALUATION CANNOT DEFEND: Val_Signal NA, FV_per_share -3.38 (negative EPV), DCF blank, Quality_Pass False, val_adj 0 - 100% signal-originated. UT-9 FOURTEENTH SESSION: PROFIT_ARMED INERT (Gain_ATR -3.09, -24.50%, arm-on-loss); T12_Progress_Pct 146.15 is the systemic-finding-3 artifact (metric = Close / Target_12Week). ADR-0020: peak_stop 111.93 = live stop, 0% erosion, did not fire. ESTIMATES (f=0.096): 1wk n/a (AVOID row prints no Target_Price) - honest band Est_Low 92.29 / Est_High 140.59 = -18.5% / +24.1%, widened sharply both ways; 4wk n/a; 12wk 77.51 = -31.58%. L8: MARKET order, a gap does not invalidate it. FALSIFIER: decisive close above EMA20 122.16 with T12 turning positive repairs the basis and AAOI re-enters as a FRESH buy candidate, subject to the quality gate it currently fails badly. · news |
| NVDA | NVIDIA Corporation | 227.98 | SELL 30% | 227.98 | 213.25 | 233.02 (+2.2%) | 280.45 (+23.0%) | 241.32 (+5.9%) | Nvidia's quarter was everything the bulls wanted - 2bn of revenue against 3bn expected, guidance far above the street, and the CEO forecasting roughly 70% growth next year versus the 45% analysts model - and the shares jumped 8.7%, rescuing a position that was 7.6% underwater the day before. That is exactly why we take a third off. In one session the price has consumed 94% of the entire twelve-week path the model projected for it, momentum is diverging from price, and only about 6% of modelled upside is left. We are not calling a top: we keep running behind a stop raised to. We are simply refusing to give back a recovery we were just handed. OPERATIONAL: ADR-0012 fired profit-take, MANDATORY DISPOSE, not an optional trim. ARMED via the T12_Progress_Pct >= 90 branch (94.47 = Close 227.98 / Target_12Week 241.32). UT-9 DOES NOT SUPPRESS: Gain_ATR +0.14 and position +0.52% - the first legitimately armed (not arm-on-loss) position this book has held. FIRED by 1 tick, tick (a): Exit_Warning = 'RSI_DIV,PROFIT_ARMED', a second distinct code on the same row. Ticks (b)(c)(d) did NOT fire - conviction ROSE; T12 241.32 > 1.02 x 227.98 = 232.54 (forward view alive but thin at +5.85%); label HOLD not NO TRADE. 1 tick -> profit_take_1 = 30%. G1 CHECKED AND DOES NOT PROTECT: the softened G1 downgrade applies only to fresh BreakoutUp_*/MomentumContinuation_* at conviction >= 80; this row is AnticipationUp_VCP. TENSION STATED: conviction 100 / Trend 94 / +8.74% gap is in spirit the strong uptrend G1 protects - resolved toward trimming because the arming basis is TARGET EXHAUSTION not gain magnitude, and because L6 independently forbids adding into a post-earnings gap, leaving only hold or trim. G2/G3 not reached (valuation-elevate path only). NO VALUATION AMPLIFY: MoS_FV -2.279 > -3, portion stays 30%, no tier bump. WHAT IS BANKED, HONESTLY: cost 226.80, so realizes just + - this is a DE-RISK, removing of exposure, not a windfall. Sizing 30% x 168 = 50.4 ->; remaining = = 2.72% of. TRAIL RAISED ON RESIDUAL 206.05 -> 213.25 (chandelier 2.0 x ATR 8.611 below the 230.47 high; above the standing stop, so valid - never lower a trail); peak_stop re-based to 213.25, ADR-0020 0% erosion. MOO reference is the settled CLOSE 227.98 - the row's Entry_Price 228.78 is a BUY-stop from the AnticipationUp_VCP signal and is not the sell-side price. Range 225.83-230.13. ESTIMATES (f=0.096): 1wk 233.02 (+2.21%), 4wk 280.45 (+22.59%), 12wk 241.32 (+5.48%) - 12wk exceeds DCF anchor 163.98. NOTE 4wk > 12wk by 16% = incoherent forward path, systemic finding 4; the arming logic used the 12wk. L4 CORRECTION: yesterday's card logged the reaction as '~-2%, muted-to-negative' from after-hours colour; the SETTLED session was +8.74%. Forward: NBIS lock-up window opens 09-08 with the residual running against the 213.25 trail. · news |
| DELL | Dell Technologies Inc | 472.26 | HOLD | 472.26 | 427.70 | 476.56 (+0.9%) | 517.06 (+9.5%) | 607.32 (+28.6%) | Dell keeps strengthening into its own earnings date. The shares made another new high, the position is now up 2.38%, and the model's twelve-week view has risen four sessions running to 28.6% above the current price. The company guided this quarter to -45bn of revenue against roughly $35bn of analyst expectation, has raised its AI-server outlook to $60bn for the year, and sits on a record 3bn order backlog. Results land 3 September, four sessions away, and the discipline is explicit: do not trim into a catalyst. We hold the full position and raise the protective stop to instead. OPERATIONAL: HOLD FULL 713, trail raised 420.84 -> 427.70 (chandelier 2.0 x ATR 22.399 below the 472.50 session high). Close 472.26 (+1.82%), new high 472.50 extending the higher-high sequence a third day. Conviction, FIFTH consecutive gain ->->76). NO TRIM BASIS ON ANY ORIGINATION CLASS: no SELL print, Exit_Warning BLANK, no PROFIT_ARMED (Gain_ATR +0.49, T12_Progress_Pct 77.76 - below the 90 arm threshold), no confirmed lower high, no stop break. Consolidation_Bullish, Trend 76, RSI 56.9. Valuation FAIR_VALUE, MoS -2.841 (> -3, no tier effect), Quality_Pass True, F 8 - amplify-only and nothing to amplify. CATALYST-AWARE HOLD BINDS: FQ2 earnings 09-03, now 4 sessions out and INSIDE the <=5-session window; no pre-trim, a trim would require a confirmed lower high or a decisive stop break. Known risk into the print is margin compression on AI-server mix; at least one sell-side downgrade landed specifically ahead of it. ADR-0020: peak_stop 455.32 vs live 427.70 = 6.07% erosion, under the 15% threshold - DID NOT FIRE, and the erosion is NARROWING as the trail climbs. The signal's own Stop_Price prints 459.13 (today's low, 0.59 ATR below close) - far tighter than the managed chandelier and whipsaw-prone into an earnings print; THE CHANDELIER GOVERNS (systemic finding 6). ESTIMATES (f=0.096): 1wk 476.56 (+0.91%), 4wk 517.06 (+9.49%), 12wk 607.32 (+28.60%) - exceeds DCF anchor 485.61. The 08-20 armed half-exit withdrawal is now vindicated by against the 433.19 low-water card. · news |
| LITE | Lumentum Holdings | 956.14 | NO ACTION | 956.14 | 918.01 | 969.59 (+1.4%) | 1096.21 (+14.6%) | 1142.66 (+19.5%) | Lumentum clears the conviction bar for a fourth straight session and is refused for a fourth straight session by the fundamental quality screen, which reads its trailing free-cash-flow yield of 0.15% and F-Score of 6 and says no. Meanwhile the stock has risen 15.2% since the block first fired. Recorded, not acted on. OPERATIONAL: adj 66 CLEARS the 65 floor by 1; Quality_Pass FALSE (F 6, Quality_Rank 22.5/31, FCF_Yield 0.149%, Earn_Yield 0.388%) blocks NEW ENTRIES only. MoS_FV -32.505 (FV/sh 24.78) -> val_adj -10, maximum haircut. Counterfactual series now 4 consecutive sessions: 830.17 -> 885.57 -> 939.03 -> 956.14 = +15.2% while blocked - the cleanest available pricing of the gate's cost and the item that should headline the next learnings review (systemic finding 1). Two independent reasons it would not have been bought regardless: signal is FailedBreakdown_Up_20D, the L1/UT-12 discounted family, and Entry_Distance_ATR 0.00 means the entry IS the close. Ledgered as a direction:none counterfactual per learnings advisory (d)(iii). |
| SNOW | Snowflake Inc. | 329.11 | NO ACTION | 329.11 | 320.80 | 331.75 (+0.8%) | 356.62 (+8.4%) | 468.98 (+42.5%) | Snowflake re-entered the buy pool with a strong print and clears the conviction bar by a single point, but it is loss-making on the screen's measure and fails the quality gate that governs new positions. Recorded, not acted on. OPERATIONAL: conviction rebounded on a +4.36% day; adj 66 clears the 65 floor by 1. Quality_Pass FALSE (F 5, Quality_Rank 20.5/31, FCF_Yield 1.01%, Earn_Yield -1.02% negative) blocks new entries. MoS_FV -1.875 -> val_adj -9. Signal FailedBreakdown_Up_20D (L1/UT-12 discounted family), Entry_Distance_ATR 0.00 = entry is the close. Ledgered as a direction:none counterfactual per learnings advisory (d)(iii). |
| NET | Cloudflare, Inc. | 308.23 | NO ACTION | 299.10 | 281.24 | 301.65 (+0.9%) | 325.65 (+8.9%) | 374.18 (+25.1%) | Cloudflare produced the second-best pullback setup in the book on an 8.2% day, but lands one point short of the conviction floor after the valuation haircut and separately fails the quality screen. Recorded, not acted on. OPERATIONAL: conviction 74 on PullbackBuy_VWAP (the L1-favoured family), Buy_Rank 6, Entry_Distance_ATR -0.51 (entry 299.10 vs close 308.23 = a genuine pullback level, would have queued rather than chased). MoS_FV -8.501 -> val_adj -10 leaves adj 64, ONE SHORT of the 65 floor. Quality_Pass FALSE (F 4, Quality_Rank 25/31, FCF_Yield 0.30%, Earn_Yield -0.18%) blocks it independently. Two failures, either sufficient. Ledgered as a direction:none counterfactual per learnings advisory (d)(iii). |
| SAP | SAP AE | 221.17 | NO ACTION | 220.53 | 203.92 | 223.72 (+1.4%) | 253.74 (+15.1%) | 313.24 (+42.0%) | SAP rallied 4.5% with the tape but its conviction score fell, leaving it well below the bar and still failing the quality screen. Recorded, not acted on. OPERATIONAL: conviction FADED despite the +4.48% day; MoS_FV -1.330 -> val_adj -7 leaves adj 56, NINE short of the 65 floor. Quality_Pass FALSE (F 8 but FCF_Yield 0.246%, Quality_Rank 19/31). RSI 71.1 now extended. The 08-27 UBS downgrade-to-neutral that overrode this name a session ago is unrefuted by today's move. Ledgered as a direction:none counterfactual per learnings advisory (d)(iii). |
Outcome & track record
Accuracy at the time of this record.
24 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 →
0.096 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.