Recommendations
1 to acquire · 3 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 |
|---|---|---|---|---|---|---|---|---|---|
| AAOI | Applied Optoelectronics, Inc. | 135.63 | BUY — QUEUED @140.68 | 140.68 | 113.84 | 136.03 (-3.3%) | 150.13 (+6.7%) | 179.34 (+27.5%) | Applied Optoelectronics just posted its fifth straight record quarter and returned to profit: Q2 revenue (+86% year-on-year), non-GAAP EPS ahead of guidance, 800G volumes more than doubling and guided to roughly five times higher in Q3, with Q3 revenue guided to -290m. The stock spiked 13% on the news, then gave back most of it the next session, closing 9.2% off its high - so rather than chase the gap we place a stop order just above that level and only buy if the market confirms the breakout. The company has no usable valuation anchor (negative earnings power value, no DCF), so this is a purely technical, momentum-driven entry sized to a defined risk. OPERATIONAL: STRONG BUY cv90 Buy_Rank 1 on BreakoutUp_20D, raw 90 clears the US-16 chase-family bar of >=85; adj 83 after the -7 BreakoutUp_20D|buy cell. Valuation NA so val_adj 0 and the quality gate is OFF. = (85.49% x 200000 x 20%) / 140.68, size_mod 1.0, cost vs. STOP IS MANUAL 113.84 (1.5 ATR from entry, risk /sh = = 3.3% of) NOT the model's 85.04 (3.11 ATR, would risk 40% of the entry) - setting a tighter manual stop than the model is permitted; US-29 only bars the reverse. HARD US-9 GUARD: if the Mon 08-10 open gaps >~1.5% past 140.68 (opens above ~142.80) DO NOT FILL - withdraw. L6 satisfied in substance: the day-1 gap already faded intraday and the entry sits above the close, so the order fills only on a confirmed reclaim (US-21/US-35 buy-STOP instrument). Target_Price 251.97 is the risk objective, distinct from the 4wk estimate; no DCF anchor exists so no exceeds-DCF annotation is possible; the 12wk band [52.65-387.82] is 7.4x wide - directional only. The 1wk point is NEGATIVE vs entry (-3.31%), which is itself the argument for the stop order over a market order. · news |
| KLAC | KLA Corporation | 198.11 | SELL 30% | 198.11 | 180.13 | 198.38 (+0.1%) | 202.77 (+2.4%) | 213.85 (+7.9%) | This is a risk-management cut, not a change of view on the business. KLA's stop-loss level has drifted 35.7% below its own high-water mark over twenty sessions while the position has never once been reduced - a risk line walking away from a full-size holding is a discipline failure regardless of what the stock does, so we take 30% off and keep the 700-share core. The company itself is doing well: Q2 revenue 66bn (+15.2%), next-quarter guidance around $4bn, earnings 5.1% ahead of consensus, and the average Wall Street target of sits 17% above today's price on the back of a 3trn Samsung/SK Hynix capex cycle. That constructive backdrop is exactly why we trim to the minimum tier rather than exiting. OPERATIONAL: ADR-0020 Trail Erosion FIRST fire, live stop 175.00 vs Peak Stop 272.00 = -35.66%, zero size reduction ever, veto-immune per US-36. DE-ESCALATED 50% -> 30% per US-32: the signal leg DISSOLVED (STRONG SELL - REDUCE cv76 -> NO TRADE cv0 on a +2.53% session), so the two-consecutive-print limb that satisfied US-18/34 on 08-07 is gone and US-17 bars a bare NO TRADE from originating; 30% is the ADR-0020 configured floor tier for us. NoTrade|sell -2 reported NOT applied (US-38). MoS -5.68 tier bump SUPPRESSED by G2 (Est_12W 213.85 = 2.56x the 83.43 DCF). ADR-0012 armed via T12_Progress 92.64 (US-26 passes: Gain_ATR 1.86, +12.72%) but NO tick fires - no 2nd warning code, cv76->cv0 is a label flip not decay, Est_12W is 1.079x close and US-33 bars tick (c) alone on a T12-armed name, only 1 NO TRADE print. Banks (+12.72%) vs the 175.75 cost. Remaining 700 RAISE stop 175.00 -> 180.13 (1.5 ATR) - first raise available in weeks; watermark re-bases to 180.13 on execution and future fires face the full veto stack. US-37 session 8. TSMC July monthly ~08-10 and AMAT FQ3 08-13 are read-throughs, not KLAC's own catalyst. · news |
| GOOGL | Alphabet, Inc.- Class A | 354.30 | SELL 30% | 354.30 | 356.00⚠ | 355.35 (+0.3%) | 359.38 (+1.4%) | 384.37 (+8.5%) | Taking a third of a position that has almost tripled, after the shares slipped below their protective stop and the model has gone quiet for three straight sessions. Alphabet's quarter was outstanding - revenue 8bn (+24.2%), Cloud accelerating 82%, earnings of against a consensus - and the market sold it anyway, which is the warning worth acting on. Underneath that, the demands keep growing: a fresh -25bn bond offering on top of a reported ~$80bn raise and capex guided to -205bn, while the AI leadership ranks keep reshuffling (Demis Hassabis moving to Alphabet chief scientist and chairman, Jeff Dean and Sanjay Ghemawat leaving to found Discovery Loop). We bank a partial gain and hold with the forward view still positive. OPERATIONAL: ADR-0012 armed + fired. ARMED via Gain_ATR 17.53 (>=3.0) with Unreal_PnL +188.05% - the independent gain limb, so US-33 does not apply and US-26 passes cleanly. FIRED on tick (d), now THREE consecutive settled NO TRADE prints (A05C/A06C/A07C cv0). Ticks (a)/(b)/(c) absent: PROFIT_ARMED is the sole warning code, cv drift zero, Est_12W 384.37 = 1.085x close vs the 1.02 dead-forward bar. This is a RE-PRICE of the unexecuted 08-07 card at the live level, NOT a US-19 escalation. STOP BREACH: close 354.30 vs the recorded 356.00 = -0.48%, MARGINAL not decisive (US-22 held MU through -0.21%/-0.71%), so the tier stays 30% rather than the 50% the 08-07 re-arm contemplated - that re-arm was written contingent on the first 30% having executed, and it did not. Valuation SILENT (MoS -0.74, far short of the -3 amplify threshold) - originates nothing per CLAUDE.md section 9. NoTrade|sell -2 reported NOT applied (US-38). Overlay's <345 GOOGL override is 34 days stale - US-31 strips its originating force. Banks (+188.05%) vs the 123.00 cost. Remaining 112 keep stop 356.00 (1.5 ATR = 334.51, no raise). RE-ARM MADE EXPLICIT: a DECISIVE settled close below 356.00 - 1.5% or more, i.e. <= 350.66 - fires a further 30% trim on the remainder; a -0.48% touch is not a break. · news |
| SPCX | Space Exploration Tech Corp | — | SELL 50% | 133.11 | 114.92 | n/a | n/a | n/a | We committed last session to halving this position unless the shares closed back above our stop level, and they finished at - short. So the trim goes through, and it is worth being straight that it sells into real strength: SpaceX rose 15.8% on Friday, its best day ever, on an Argus upgrade to Buy with a target, ended the week up 23%, and absorbed a very large lock-up expiry without damage. The reason we honour the line anyway is that this same card has been written and not acted on three sessions running, and drifting past our own pre-set levels has cost this book more than any analytical call. We keep half the position and re-set the stop to the level Friday's move launched from. OPERATIONAL: hardened date-bound re-arm from the EOD 2026-08-07 card fires as written - 'MANDATORY 50% DISPOSE CARD at the 08-10 EOD unless it posts a settled close above 135.00 between now and then'; this run IS that adjudication (the scheduler's Mon 08-10 06:00 GMT+8 EOD anchors this same Fri 08-07 ET close). Close 133.11 = -1.40% below the line, not cleared. PRICE IS NEWS-SOURCED (Fri 08-07 settled close) - SPCX has no scan row, no Entry_Price, no Est_* columns, no conviction, no ATR; permanently muted, private-market listing, nothing synthesized. Cost is unrecorded in usportfolio.json so realised cannot be stated. The unlock leg (911.5m shares, 3.3x float, ~$100bn) resolved outright on 08-06; not executing the 08-06 card at 108.27 is +ahead. Valuation NA (EPV -1.90, no DCF) - silent. Overlay SPCX override 34 days stale (US-31). Remaining 80: RE-BASE the manual stop 135.00 -> 114.92 (the 08-06 settled close, the base the +15.83% session launched from). AUTO-UNMUTE and resume full signal+fusion treatment the moment SPCX appears in the scan CSV with a non-null Close. US-37 session 3. RKLB Q2 08-10 is the closest sector read-through. · news |
| NVDA | NVIDIA Corporation | 223.96 | HOLD | 223.96 | 212.52 | 225.89 (+0.9%) | 227.28 (+1.5%) | 238.17 (+6.3%) | The strongest holding in the book this session - conviction rose sharply on a 2.3% gain, and the shares sit comfortably above every major moving average with earnings still three weeks out. We simply raise the protective stop and hold. OPERATIONAL: HOLD cv68 -> cv88, conviction RISING 20 points on AnticipationUp_VCP, +2.27% to 223.96, above EMA20 207.89 / EMA50 205.80 / EMA200 194.84, RSI 64.4. PROFIT_ARMED prints for a 5th session but US-26 DECLINES the arm (no Cost in usportfolio.json -> Gain_ATR and Unreal_PnL_Pct blank, gain unverifiable; systemic #6). A 20-point conviction RISE is the opposite of an ADR-0012 deterioration tick. Stop 206.93 -> 212.52 (1.5 ATR below the close). Earnings 08-26; CRWV 08-11 and COHR 08-12 are read-throughs. SHARED WITH ussemicon - execute once (L7). |
| NET | Cloudflare, Inc. | 300.27 | HOLD | 300.27 | 272.00 | 298.90 (-0.5%) | 308.34 (+2.7%) | 328.90 (+9.5%) | Cloudflare's Q2 beat on both revenue and earnings with full-year guidance raised, and the shares added 5.6% to close at, putting the position 16.4% ahead of cost. We hold all but do not add: this is only the second day of the post-earnings move and the model's own buy trigger sits 3.3% above the current price, so adding here means chasing the gap rather than buying a pullback - a pattern that has repeatedly round-tripped in this book. If the shares consolidate back to the trigger level with the signal intact, the add re-opens. OPERATIONAL: NO TRADE cv0 -> STRONG BUY cv90 on MomentumContinuation_Up, Buy_Rank 2. adj = 90 - 10 (MoS -943.71, an EPV artifact - FV_per_share prints) - 10 (MomentumContinuation_Up|buy, the worst actionable cell, n=6 hit.0) = 70, five over the floor; the quality gate cannot block it because NET is HELD so this would be an ADD not a new entry. DECLINED ON L6 - day 2 of a post-earnings move, buy-stop 310.03 sits 3.25% ABOVE the 300.27 close so there is no pullback to buy. Above all four EMAs (EMA20 276.74 / EMA50 258.34 / EMA200 224.31), +16.39% (+) over the 258.00 cost. Stop 272.00 STANDS (1.5 ATR = 269.18; model 247.86 is LOOSER - REJECTED per US-29). WARNING 8th SESSION: two NET rows in input/usportfolio.json ( and), the loader takes the LAST, so this position is invisible to the scan (Avg_Cost blank, Held false) and reads to the monitor as unledgered_drift. DE-DUPLICATE BEFORE THE NEXT RUN. · news |
| NBIS | Nebius Group, N.V. | 187.97 | HOLD | 187.97 | 183.00 | 194.62 (+3.5%) | 209.24 (+11.3%) | 269.17 (+43.2%) | The single bearish reading that appeared two sessions ago has faded back to neutral, the stop is still intact with 2.7% of room, and results land pre-market in two sessions - so we hold into the print rather than trimming ahead of it. A pre-set trim remains armed if the stop breaks or the bearish signal returns and holds. OPERATIONAL: the 08-06 lone STRONG SELL - REDUCE DISSOLVED to AVOID cv59, confirming US-18/34's bar was right to hold it. Close 187.97 (-1.01%), stop 183.00 intact by +2.72%. NOT armed (Gain_ATR -0.97, -11.17% - US-26 declines outright). adj = 59 + 1 (FailedBreakout_Down_20D|sell) = 60, five under the floor, and AVOID is not an actionable sell label in any case. Stop_Price prints 187.97 = Close - the US-29 artifact, REJECTED. 1.5 ATR = 151.56, no raise. CATALYST-AWARE HOLD: Q2 pre-open 08-12, two sessions out, CRWV 08-11 the closest read-through - no pre-trim into the print. PRE-COMMITMENT CARRIED: any settled close below 183.00, OR a 2nd consecutive settled STRONG SELL - REDUCE, fires a mandatory 50% trim - NBIS has never been size-cut. |
| MU | Micron Technology, Inc. | 877.57 | HOLD | 877.57 | 770.00 | 882.32 (+0.5%) | 977.55 (+11.4%) | 1055.12 (+20.2%) | An eighth straight session with no actionable signal - the model has nothing to say about this name either way. The position is modestly ahead of cost and far above its long-term trend, so we hold and leave the stop where it is. OPERATIONAL: 8TH consecutive NO TRADE cv0 (US-17 bars origination on a bare NO TRADE label), NoTrade_Reason verbatim 'Consensus conflict + low conviction'. -0.44% to 877.57, +2.10% over the 859.49 cost. Under EMA20 895.21 / EMA50 889.51, far above EMA200 608.73. NOT armed (T12_Progress 83.17, Gain_ATR 0.23). 1.5 ATR = 761.22 - below the standing 770.00, no raise. 12wk band [470.82-2140.93] is 4.6x wide - weight the 1wk (882.32, +0.54%). Re-based Trail Erosion remains VETOED (US-17, US-30: above cost and above the line). TSMC July monthly ~08-10, AMAT FQ3 08-13. |
| OKTA | Okta Inc. | 148.32 | HOLD | 148.32 | 140.00 | 148.84 (+0.4%) | 152.79 (+3.0%) | 169.22 (+14.1%) | Up 3% on the session and now 6.1% ahead of cost, with the protective stop already sitting above the purchase price - the position is stopped into a profit and needs no action. OPERATIONAL: HOLD cv64 -> cv56 on FailedBreakdown_Up_20D, +2.97% to 148.32, +6.09% over the 139.80 cost, above EMA20 141.61, well above EMA50 130.34 / EMA200 104.16, RSI 60.9. NOT armed (Gain_ATR 1.31, T12_Progress 87.65). Stop 140.00 STANDS: the 1.5 ATR chandelier is 138.59 (looser), and the model prints Stop_Price 145.08 which is TIGHTER at only 0.50 ATR - noted but NOT adopted, whipsaw risk on a 6.49 ATR name. WATCH: a 2nd consecutive NO TRADE would arm ADR-0012 tick (d). |
| DELL | Dell Technologies Inc | 453.77 | WATCH | 453.77 | 407.00 | 456.34 (+0.6%) | 493.11 (+8.7%) | 606.89 (+33.7%) | An unresolved bookkeeping item, not a trading decision: a buy order placed on 08-05 should have filled, but no Dell position appears in the portfolio file and no transaction was recorded. It needs to be marked as either taken or skipped before the next run, because an unrecorded holding is invisible to every risk check. On the signal itself there is nothing to act on today. OPERATIONAL: ADR-0019/US-37 session 3. The 08-05 buy-STOP at 464.24 (expires_on 2026-08-10) almost certainly FILLED - DELL opened 466.60 on 08-06, above the trigger, with no gap, so it fills at the open by construction. If held, are now marked 453.77 = -2.75%, - (improved from -last session). Record TAKEN (trade buy) or SKIPPED (trade skip us DELL). NO RE-QUEUE: tonight's BUY cv69 grades adj = 69 - 10 - 7 = 52, thirteen under the floor. NOTE the -10 val_adj is a plumbing artifact - DELL is FAIR_VALUE, Quality_Pass True, F-Score 8, DCF 485.61 vs a 453.77 close (genuinely cheap on DCF) but FV_method EPV prints 112.78, driving MoS_FV -2.88; see systemic finding #2. Stop 407.00 stands if the position is held. |
Outcome & track record
Accuracy at the time of this record.
121 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.