💼 Model Portfolio - Auto Run by Agents — Oct 14, 2026

💼 MODEL PORTFOLIO — Auto Run by Agents — 2026-10-14

Cumulative P&L +10.63% | Daily P&L -1.03% | Gross 80% | Net +0% | Net Beta +0.16 | Vol(ann.) 6.6%

Book d-1.03% c+10.63% vs SMH d-3.98% c-3.78% | α(d)+2.94% α(c)+14.41% · SOXX d-4.50% c-6.50% | α(d)+3.47% α(c)+17.13%

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

Market: Thu Oct 15 8am SG run = processing the Wed Oct 14 US session. Tape: AI-CAPIEX ROUT - CRWV -15.46% (bond ~15% yield debt panic), NBIS -10.08%, AAOI -11.82%, DELL -11.27% (Barclays EW on enterprise-AI timing), CIEN -10.19%, CLS/FLEX -9.5%, ORCL -8.83% (financing theme) + CPI; ASML -3.94% DESPITE beat/raise (Q4 EUR14.2-15.2B, FY26 EUR44-45B, 2027/28 output won't meet demand, 2029/30 PO requests, value-based pricing done) = sell-the-news. GFS +4.80% top gainer (two new NY fabs, U.S.-Japan initiative). Day -1.03%; Cumulative +11.79%->+10.63%; Gross 80.0% (full replenishment), Net 0.0% EXACT neutral. TRADED: CLOSED MU (-12.44% hard stop through -10%), CLOSED VICR (-13.35% hard stop + armed into Oct 20), CLOSED GFS (bank-and-CLOSE +17.32% - legs inverted, new NY fabs, 50d reclaimed), TRIMMED DELL 3.0->2.0 (Barclays mechanism counter-catalyst), TRIMMED ARM 2.5->2.0 (framework #3 corrected basis +26.29%); OPENED AMAT long 2.5 (ASML demand confirmation; re-entry), AAPL long 2.5 (BNP upgrade + HomeView), ORCL short 2.0 (AI-infra financing mechanism), SPCX short 2.0 (neocloud financing family). BASIS REPAIR: all short marks recomputed to (entry-cur)/entry; NXPI Oct 13 trim re-underwritten (fired on covering-basis inflation +23.12% vs correct +18.78%; trim stands, second trim deferred to framework #4 into Oct 27). Watch: NFLX -5.23% (Oct 20 binary, 2 sessions), TSEM -5.87% (optical contagion day 2), CRWD -5.03% band zone, PANW -4.64%, NOW -4.17%, ADBE -2.21%, CRM +0.69% (close if >$242), ORCL short squeeze-exit clause into TSM TONIGHT. Forward: TSM Q3 Oct 15 AMC (tonight - the named squeeze trigger for the ORCL short), NFLX Oct 20, META/IBM/SAP/TXN/LRCX Oct 21, BE/NOK/SPOT/WDC Oct 22, AMKR/CLS/CDNS Oct 26, GLW/TER/NXPI Oct 27, MSFT/GOOGL/NOW/FTNT/ASMIY/UMC/FLEX/GEV Oct 28, AMZN/INTC/ENTG/MPWR/STM Oct 29 + AMD DevDay, AAPL/CHKP/TSEM Nov 2, AMD/ANET/SMCI/ALAB/SPCX Nov 3, ARM/QCOM/MKSI/Q Nov 4, AMAT Nov 12 (est), NVDA Nov 17/18, DELL/ADI/ZS Nov 24. | PRIOR (Wed Oct 14 8am SG run = processing the Tue Oct 13 US session. Tape: semi-equipment ROUT on company news (AMAT -6.69% $5B EPIC Center w/o 2030 target, ONTO/KLAC/LRCX/TER all -2.7% to -3.7%, CDNS -4.08%) vs quality/AI-infra bid (ORCL +4.23% Project Jupiter, META +2.63%, TSEM +2.63% Tower-Semtech SiGe deal, SMTC +2.67%); memory -3.6% (MU -3.62%, SKHY -3.19%) on CXMT DRAM-densification jitters. Day +0.15%; Cumulative +11.61%->+11.82%; Gross 80.0% (floor, in band); Net +2.0%. TRADED: CLOSED AMAT long (hard-armed -9.33% into ASML Oct 14 binary + EPIC counter-catalyst + AMAT/ASMIY pair failing); OPENED TSEM long 2.5% (replacement mandate, Tower/Semtech SiGe AI-interconnect deal). Watch: MU -8.70% HARD-ARMED (HOLD per VICR, no binary till Dec 23), CRM short -0.18% (close >$242), NFLX -4.83% (Oct 20), NOW -3.05% (Oct 28). Forward: ASML Q3 Oct 14 (WFE binary, ASMIY short exposed), TSM Oct 15, NFLX/VICR Oct 20, IBM Oct 21, STX/NXPI/SKHY/GLW/ASMIY Oct 27, big-cap Oct 28 (MSFT/GOOGL/META/NOW), AMZN/INTC Oct 29 + AMD DevDay, TSEM Nov 2, AMD/ANET/SMCI Nov 3, ARM/QCOM/GFS Nov 4, NVDA/PANW Nov 18.)

📜 Recent Trades (last 15 of 110)

• 2026-10-15 DAILY REBALANCE N/A @ None [0.0%] — Wed Oct 14 session processed (run Thu Oct 15). BROAD AI-CAPIEX RISK-OFF ROUT: CRWV -15.46% (bond yield ~15% on the 9.625% coupon = debt-sustainability panic; Chamath chart; $52B debt load), NBIS -10.08%, AAOI -11.82%, DELL -11.27% (Barclays downgrade to Equalweight on enterprise-AI TIMING; only ~25% of enterprise AI spend to hardware), CIEN -10.19%, CLS -9.55%/FLEX -9.51%/SMCI -8.42% (same note on ODMs), ORCL -8.83% (financing-cost theme), LITE -9.27% (optical contagion) + CPI print. Quality held better (AAPL +0.50%, MSFT -1.80%). GFS +4.80% TOP GAINER (Oct 14: selected to operate two new NY fabs under the U.S.-Japan initiative, on top of the Oct 8 $2B TSMC interposer deal + Marvell Vermont expansion). ASML -3.94% DESPITE beat+massive raise (Q4 EUR14.2-15.2B, FY26 EUR44-45B, 2027/28 output 'will not fully meet demand', 2029/2030 PO requests, value-based pricing done) = sell-the-news on the debt/CPI tape. Our longs bled (DELL -11.27%, VICR -8.19%, TSEM -5.87%, AMD -5.73%, PANW -5.54%, CRWD -4.56%, ANET -4.45%, MU -4.10%, TSM -3.40%, NVDA -3.16%, ADI -2.68%, MSFT -1.80%, IBM -1.26%) vs shorts paid (ARM -6.81%, SKHY -6.06%, SNDK -5.33%, GLW -5.15%, UMC -4.42%, WDC -4.36%, CAT -3.08%, ASMIY -2.94%, QCOM -2.40%, NXPI -2.43%, AMZN -1.74%, CRM -0.87%); gave back GFS +4.80%, ADBE +1.86%, NOW +0.99%, NFLX +0.16%. Day -1.03%; Cumulative +11.79%->+10.63% (compounded). TRADED: CLOSED MU (HARD STOP -12.44% through the -10% band; the Oct 13 armed-hold plan said 'let the -10% stop work' - it worked), CLOSED VICR (HARD STOP -13.35%, armed into its own Oct 20 print = double reason), CLOSED GFS (bank-and-CLOSE +17.32%: NVMI/UMC family - 'no-AI-exposure' leg killed twice over, 2 new NY fabs = utilization leg transforming, +4.80% against the tape on own news, reclaimed 50d), TRIMMED DELL 3.0->2.0 (Barclays mechanism-level counter-catalyst + 50d break), TRIMMED ARM 2.5->2.0 (framework #3 on corrected basis +26.29%); per replacement mandate OPENED AMAT long 2.5 (ASML demand confirmation, re-entry re-underwriting the Oct 14 close), AAPL long 2.5 (BNP upgrade + HomeView), ORCL short 2.0 (AI-infra financing-cost mechanism, Big Tech pair), SPCX short 2.0 (neocloud financing family, contained chase). Gross 80.0->80.0% (full replenishment), Net +2.0->0.0% (EXACT neutral). EVALUATED-AND-LEFT: SPOT long 4th REJECT ('revisit on pullback' trigger still unfired: -0.66% drift, still ~+9% off Oct 6 = relative strength in the rout, no pullback; NFLX pair slot stays empty), CRWV/NBIS shorts (single-day chases -15.46%/-10.08%; SPCX chosen as the contained expression), CLS/FLEX shorts (chase -9.55%/-9.51% on the same Barclays note + CLS Oct 26 print/Oct 27 Investor Day event risk + SMCI retired-thesis trap), ON/MPWR shorts (technical-only 50d breaks, no company catalyst - 4th+ reject), GFS long flip (+4.80% chase, catalyst priced on the day), UMC add (GFS US mature-node fab wins mildly counter the 'no demand' legs; Foundry stays +1.0 net documented), optical shorts CIEN/AAOI/LITE (-9 to -12% chases on a macro funding narrative). Watch: NFLX short -5.23% into Oct 20 binary (2 sessions), TSEM -5.87% (day 2 of optical contagion), CRWD -5.03% band zone, PANW -4.64%, NOW -4.17%, ADBE -2.21%, CRM short +0.69% (close if reclaims/holds >$242), ORCL short pre-committed squeeze exit (TSM tonight). Forward: TSM Q3 TONIGHT Oct 15 AMC (the AI-capex squeeze trigger for the ORCL short), NFLX Oct 20, META/IBM/SAP/TXN/LRCX Oct 21, BE/NOK/SPOT/WDC Oct 22, AMKR/CLS/CDNS Oct 26, GLW/TER/NXPI Oct 27, MSFT/GOOGL/NOW/FTNT/ASMIY/UMC/FLEX/GEV Oct 28, AMZN/INTC/ENTG/MPWR/STM Oct 29 + AMD DevDay, AAPL/ON/CHKP/TSEM Nov 2, AMD/ANET/SMCI/ALAB/SPCX Nov 3, ARM/QCOM/MKSI/Q Nov 4, AMAT Nov 12 (est), NVDA Nov 17/18, DELL/ADI/ZS Nov 24.
• 2026-10-15 P&L BASIS REPAIR N/A @ None [0.0%] — MANDATORY REPAIR per references/price-fetch-and-pnl-math.md: every short's unrealized_pnl_pct recomputed on the correct simple basis (entry-cur)/entry*100, killing the covering-return inflation re-introduced on the Oct 13 AND Oct 14 runs (stored Oct 13 marks were NXPI +22.52 vs correct +18.38, SNDK +27.30 vs +21.45, GFS +26.75 vs +21.14, ARM +26.43 vs +20.93). Corrected Oct 14 marks: NXPI +20.36, AMZN +4.94, ARM +26.29, SNDK +25.63, CRM +0.69, UMC +13.29, WDC +23.14, ADBE -2.21, QCOM +7.93, NOW -4.17, NFLX -5.23, GLW +11.60, CAT +7.97, ASMIY +10.50, SKHY +9.05 (NXPI basis assert passed before save; all shorts verified post-save). NXPI TRIM RE-UNDERWRITING: the Oct 13 framework-#3 trim (3.5->2.5 at $225.21) fired on the inflated +23.12% when the correct figure was +18.78% - BELOW the +20% trigger. Verdict: the trim STANDS (realized portion priced at $225.21 vs $220.83 today = it outperformed holding) but was premature by framework math; the +20% trigger is only genuinely met NOW (+20.36%) and the second trim is DEFERRED to the framework #4 pre-print trim into the Oct 27 print to avoid double-banking the same framework twice in two sessions. All prior trims audited: GFS Oct 9 (+16.61% correct basis) and the Oct 7 bank trims predate the bug - clean. Rationale cum figures rewritten on the corrected basis this run.
• 2026-10-15 CLOSED LONG MU @ 914.13 [3.5%] — HARD STOP - closed at $914.13, realized -12.44% on 3.5% = -0.44% portfolio (entry $1,043.96, Sep 21). THROUGH the -10% hard band (-8.69% -> -12.44% in one session, -4.10% day on the CXMT DRAM-densification unwind). The Oct 13 armed-hold plan was explicit: 'HOLD per armed-hold precedent (let the -10% stop work)' - the stop worked, and the band is the band (CHKP -10.72%/TXN -10.23%/MPWR -10.86% precedent: a breach past -10% closes regardless of the HBM-sold-out/SCA leg). No binary until Dec 23 = no catalyst to hold into. Memory book flips to pure short family (SNDK 2.5/WDC 1.5/SKHY 2.0) - deliberate net short = the CXMT supply thesis acts as a book-level hedge vs AI longs.
• 2026-10-15 CLOSED LONG VICR @ 245.35 [1.5%] — HARD STOP - closed at $245.35, realized -13.35% on 1.5% = -0.20% portfolio (entry $283.16, Sep 23). THROUGH the -10% hard band after a -8.19% capitulation day (267.24->245.35) that erased the Oct 13 USITC-patent relief rally. DOUBLE exit reason: past the band AND armed into its own Oct 20 print (2 sessions) - the SMCI/STX 'don't hold an armed/inverted thesis into its own event' family. The analog/power rotation resumed day 6; thesis (AI OEM licensing/ChiP ramp) not disproven - 're-entry candidate post-print or on rotation stabilization'.
• 2026-10-15 CLOSED SHORT GFS @ 48.94 [2.5%] — BANK-AND-CLOSE (NVMI Sep 4 / UMC Aug 25 family) - closed at $48.94, realized +17.32% on 2.5% = +0.43% portfolio (entry $59.19, Jul 22; post-trim core banked). 2 of 3 entry legs inverted and the third transformed: (1) 'no AI exposure' killed twice over (Oct 8 $2B TSMC CoWoS interposer deal + Marvell Vermont AI optical-interconnect expansion), (2) 'low utilization' being actively fixed by the Oct 14 selection to operate TWO NEW New York fabs under the landmark U.S.-Japan initiative (fresh company catalyst dated Oct 14), (3) mature-node pricing pressure unverifiable against a capacity-expansion mandate. Tape verdict: +4.80% TOP GAINER against a -1% tech rout on its own news + reclaimed the 50d (ma50 file) = the laggard has become the US-manufacturing/AI-packaging narrative. Precedents: UMC (laggard becomes the AI narrative -> close even with the old thesis intact), QCOM Sep 8 (customer deals re-rate the short). Remaining trade is mean-reversion on a re-rated name - not ours.
• 2026-10-15 TRIMMED LONG DELL @ 507.93 [1.0%] — Trim 3.0->2.0 at $507.93 - fresh opposing marquee-analyst counter-catalyst (CRM Oct 13 family): Barclays downgrade to Equalweight from Overweight citing enterprise-AI TIMING ('only ~25% of enterprise AI spend goes to hardware') = a mechanism-level claim against backlog conversion, plus a 50d break (42 days above) on an -11.27% day. Realized +14.66% on the trimmed 1.0% = +0.15% portfolio. Core 2.0% stays for the Nov 24 print: the $95B backlog/$60.9B orders are contracted demand, not refuted by a spend-MIX forecast - one analyst at thick cushion would have been a HOLD (DELL Sep 11 rule) but the mechanism claim + technical break justify banking 1/3 (UMC 1-leg-inverted trim precedent).
• 2026-10-15 TRIMMED SHORT ARM @ 238.01 [0.5%] — Trim 2.5->2.0 at $238.01 - framework #3 bank-gains on a corrected-basis +26.29% winner (entry $322.90, Sep 21; never banked since entry; ANET 4->3 at +22.49% / GFS 22.4% precedent). The AI-CPU-premium deflation is paying on schedule (-6.81% today). Realized +26.29% on the trimmed 0.5% = +0.13% portfolio. Core 2.0% rides into the Nov 4 print with framework #4 pre-print trim discipline next (GFS multi-trim sequence).
• 2026-10-15 OPENED LONG AMAT @ 446.63 [2.5%] — RE-ENTRY long 2.5% at $446.63 (Oct 14 close) - replacement mandate for the MU/VICR/GFS book + NEW DISTINCT CATALYST reversing the Oct 14 risk-close (QCOM Sep 29 re-entry family; the close reason explicitly kept the thesis alive): ASML Q3 (reported Oct 14 pre-mkt) beat and RAISED across the board - Q4 guide EUR14.2-15.2B, FY26 raised to EUR44-45B, planned 2027 AND 2028 output 'will not fully meet demand' with the 2028 expansion investigation ABOVE +30%, customers requesting POs for 2029/2030 (NEW), value-based pricing conversations COMPLETED (price increases secured, full effect 2028) = the WFE demand/pricing backdrop the AI-capex/WFE-leader thesis needed. Re-underwriting the close: (a) the ASML binary that made holding an armed long asymmetric RESOLVED BENIGNLY; (b) the EPIC counter-catalyst ('$5B center without the 2030 target') is a nit against an industry sold out through 2028-30; (c) the AMAT/ASMIY pair failure stays fixed - AMAT is NOT re-paired (ASMIY short kept standalone; book-level WFE exposure documented). Entry discipline: stock -3.79% (464.24->446.63) with the tape ignoring the ASML print (CPI + CRWV debt panic) = buying the pullback on a demand catalyst, no gap-chase. Nov 12 Q3 (est) = defined event risk. Sized 2.5% (re-entry norm). Risks: WFE momentum hostile; TSM print tonight could gap the complex.
• 2026-10-15 OPENED LONG AAPL @ 342.96 [2.5%] — NEW - fresh analyst + company catalysts on a zero-chase tape: BNP Paribas upgrade to Outperform + HomeView smart-home initiative announced at the Oct 13-14 home-products event + defensive-bid relative strength (+0.50% in a -1% tech rout while DELL -11.27%, ORCL -8.83%). Cross-checks: the Oct 13 'AAPL media event (home products)' event-catalyst row + iPhone Duo preorders Oct 16 = near-term product-cycle legs. Book-level role: offset to the structural QCOM short (iPhone-cycle proxy vs the handset BOM squeeze; the Oct 14 SWKS note explicitly framed handset chip demand as steady). Big Tech pair balance: L MSFT 3.5 + AAPL 2.5 vs S AMZN 3.5 + ORCL 2.0 = L6.0/S5.5 (+0.5 net). Nov 2 Q3 print = defined event risk (framework #4 applies). Sized 2.5% (analyst+initiative tier, IBM Oct 9 precedent).
• 2026-10-15 OPENED SHORT ORCL @ 130.55 [2.0%] — NEW - AI-infrastructure FINANCING-COST mechanism, passed the cross-sector narrative test (CRM Sep 23 family): (a) credible bellwether = CRWV bonds trading ~15% yield vs the 9.625% coupon (Oct 14 debt-sustainability panic) = a sector cost-of-capital shock for debt-funded AI capex, corroborated by the Barclays enterprise-AI caution; (b) specific mechanism = rising financing costs pressuring negative-FCF capex builders; (c) purest watchlist expression = ORCL (negative FCF, FY27 capex >=$90B ABOVE estimates, $664B RPO requiring debt-funded buildout - the exact legs flagged in the Sep 10 beat-and-raise de-ration); (d) sized as a NEW thesis 2.0% (financing/capital-markets tier, GLW Oct 5 precedent). Tape confirmation: -8.83% vs Big Tech -0.5%/-1.8% = ~5x same-day underperformance of its own sub-sector (the GLW relative-weakness test). PRE-COMMITTED EXIT CLAUSE (STX family): the Sep 10->11 flip showed financing narratives whipsaw on demand data - trim/close if an AI-capex melt-up squeeze hits; TSM Q3 TONIGHT is the named trigger. Dec 9/14 print. Pair: AAPL/MSFT longs vs AMZN/ORCL shorts = Big Tech near-neutral.
• 2026-10-15 OPENED SHORT SPCX @ 152.16 [2.0%] — NEW - neocloud financing-cost family, the CONTAINED-CHASE expression of the Oct 14 CRWV debt panic: CRWV -15.46% and NBIS -10.08% were evaluated and rejected as single-day chases (STX Oct 2 discipline: a short into a -10%+ day whose legs can invert on a bounce); SPCX fell only -2.80% after its own Oct 13 -4.21% spectrum/debt-concerns slide = the own-balance-sheet leg is already the thesis, with the CRWV ~15% bond yield re-pricing the whole neocloud funding complex as the fresh sector catalyst. Mechanism: neoclouds fund AI capex with high-cost debt; a sector cost-of-capital shock hits the weakest balance sheets first. Nov 3 print (est) = defined event risk. Sized 2.0 BELOW the 2.5 norm on chase risk (SKHY Oct 13 precedent). Risks: AI-capex melt-up squeeze (TSM tonight = trigger to watch), ADR/liquidity.
• 2026-10-14 CLOSED LONG AMAT @ 464.24 [2.5%] — RISK CLOSE - realized -9.33% on 2.5% (entry $512.01, Sep 30) = -0.23% portfolio. AMAT HARD-ARMED at -9.33% (0.67% from the -10% hard band) AND (a) fresh company counter-catalyst: AMAT launched a $5B EPIC Center (Equipment Process Innovation Center) at SEMICON West Oct 13 WITHOUT the full 2030 revenue target the Street wanted -> -6.69% profit-taking (semi-equip dominated the loser board: ONTO -3.72%, KLAC -3.22%, LRCX -3.38%, TER -2.73%); (b) ASML Q3 TOMORROW (Oct 14) = WFE binary that could gap the armed long through the hard stop; (c) the AMAT long / ASMIY short WFE relative-value pair is FAILING - AMAT -6.69% UNDERPERFORMED its own short leg = long side losing. Discipline: close an armed long into a binary rather than risk a gap through the -10% stop (SMCI/STX 'don't hold an inverted/armed thesis into its own event' family). Thesis (AI-capex/WFE leader) not fundamentally broken (EPIC = long-term investment) but the RV leg is dead + near-term gap risk is asymmetric. ASMIY short kept (own FY-guidance-miss thesis intact, +8.45%, paying as WFE de-rates).
• 2026-10-14 OPENED LONG TSEM @ 222.2 [2.5%] — NEW - replacement mandate for the AMAT close + fresh Oct 13 company/contract catalysts (rank ~1-2, contract/deal family): Tower Semiconductor + Semtech EXPANDED multi-year agreement (announced Oct 13 06:00 EDT) securing dedicated SiGe capacity across Tower's US/Israel/Japan fabs to support Semtech's AI data-center optical/copper interconnect design wins (800G->1.6T->3.2T roadmap), on top of NewPhotonics high-volume laser-integrated PIC shipments (Sep 17) + the NVIDIA 1.6T silicon-photonics collaboration (Feb). TSEM BOUNCED +2.63% to $222.20 on the deal = company catalyst overcoming the optical/foundry tape (decoupling from the falling knife). The prior TSEM rejects were 'falling knife + NO company catalyst' - the Oct 13 Semtech deal REMOVES that objection (fresh contract = fundamental floor). Foundry RV long vs mature-node shorts (GFS/UMC): TSEM carries AI-photonics/SiGe upside while GFS/UMC bear mature-node pricing pressure. Entered on a modest +2.63% (not a gap-chase) after a -5% pullback (228->216->222) = better entry than the Oct 6 peak. Nov 2 Q3 print = defined event risk (framework #4 trim into it). Sized 2.5%. Risks: optical-complex contagion if AI-capex de-risk resumes; TSEM is mid-cap/volatile.
• 2026-10-14 DAILY REBALANCE N/A @ None [0.0%] — Tue Oct 13 session processed (run Wed Oct 14). Tape: SEMI-EQUIPMENT ROUT on company news while quality/AI-infra bid - AMAT -6.69% ($5B EPIC Center launch without full 2030 revenue target -> profit-taking; SEMICON West investor breakfast), ONTO -3.72%, KLAC -3.22%, LRCX -3.38%, TER -2.73%; CDNS -4.08% (pre-earnings chip-equip selloff); SPCX -4.21% (spectrum/debt concerns). Memory slid again on CXMT DRAM-densification jitters (MU -3.62%, SKHY -3.19%). Offset: Big Tech/AI-infra bid (ORCL +4.23% Project Jupiter AI infra + power approvals, META +2.63% Muse Wonder), Foundry/Optical deal flow (TSEM +2.63% Tower-Semtech SiGe deal, SMTC +2.67% same deal). Our longs: AMAT -6.69% (the loser), MU -3.62% (WATCH), ANET -3.02%, PANW -1.68%, CRWD -0.68%, DELL -0.74%; bid AMD +1.90%, ADI +2.08%, VICR +3.02% (USITC patent investigation + rising ests; relieved from hard-armed), IBM +1.66%, NVDA +0.97%, MSFT +0.42%, TSM +0.17%. Shorts paid: ASMIY -6.17% (WFE), CAT -3.33%, GFS -3.39%, SKHY -3.19%, MU-side WDC -1.61%, SNDK -0.34%, AMZN -0.37%, UMC -0.71%, NOW -1.22%, ADBE -0.78%, NFLX -0.21%; gave back CRM +1.75% (BofA-upgrade follow-through), QCOM +1.39%, NXPI +0.49%, ARM +0.06%, GLW +0.86%. Day +0.15%; Cumulative +11.61%->+11.82% (compounded). TRADED: CLOSED AMAT long (hard-armed -9.33% into ASML binary + EPIC counter-catalyst + AMAT/ASMIY pair failing); OPENED TSEM long 2.5% (replacement mandate - Tower/Semtech SiGe AI-interconnect deal, Foundry RV). Gross 80.0->80.0% (AMAT close replaced 1:1 by TSEM; in 80-120% band at floor), Net +2.0->+2.0%. MEMORY note: MU -8.70% HARD-ARMED (1.3% from -10%) on CXMT but HBM-sold-out/SCA leg intact + no binary until Dec 23 = HOLD per VICR armed-hold precedent (let -10% stop work); net memory book L3.5/S6.0 = short CXMT exposure is a net hedge. Watch: MU -8.70% HARD-ARMED, CRM short -0.18% (close if reclaims/holds >$242), NFLX short -4.83% (Oct 20 binary), NOW short -3.05% (Oct 28), ADBE short -0.34%, VICR -5.62% (relieved, Oct 20). Forward: ASML Q3 TOMORROW Oct 14 (WFE binary - ASMIY short 2.5% exposed), TSM Q3 Oct 15 (long 3.5%), SMTC Analyst Day Oct 15, NFLX/VICR Oct 20, IBM Oct 21, SPOT/WDC/NOK Oct 22, CDNS/AMKR/CLS Oct 26, STX/GLW/NXPI/SKHY/ASMIY Oct 27, MSFT/GOOGL/META/KLAC/NOW/FTNT/CAT/GEV/FLEX/UMC Oct 28, AMZN/INTC/SNDK/MPWR/STM/ENTG Oct 29 + AMD AI DevDay, TSEM/ON/AAPL/CHKP Nov 2, AMD/ANET/ALAB/SMCI Nov 3, ARM/QCOM/COHR/GFS/MKSI Nov 4, AMAT Nov 12, PANW/NVDA Nov 18.
• 2026-10-13 OPENED SHORT SKHY @ 162.23 [2.0%] — NEW - replacement mandate for the CRM/NXPI trims + fresh supply-side catalyst: CXMT (ChangXin) DRAM-capacity progress (the Oct 12 memory-selloff driver) extends the memory-supercycle-peak family that has paid SNDK +26.87% / WDC +22.44%; tape-confirmed -4.02% Monday and -11.89% in the Oct 1-8 week (weekly-gainers). Sized 2.0% BELOW the 2.5 norm on chase-risk (STX discipline) after the double-digit weekly giveback; Oct 27 print = defined event risk (framework #4 trim applies into it). Memory category now L3.5 (MU HBM long) vs S6.0 (SNDK 2.5/WDC 1.5/SKHY 2.0) = net -2.5 - book-level offset vs AI/Accelerators +4.0 long and the structural Mobile short (QCOM). Risks: memory oversold-bounce + ADR borrow (NVMI no-shortable precedent). Fallback if unshortable: skip, gross sits 78.0% as stated exception pending next catalyst.

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

Positions by Sub-Sector

---

---

Trade Thesis

• NVDA (L): OCT 14: -3.16% to $221.30 (cum +3.82%); broke 50d (28d above) on the AI-capex debt-panic tape. Nov 17/18 print. Hold 4.0% max. | *AI price hikes on 2027 Vera Rubin/Grace Blackwell; Blackwell Ultra ramp*
• TSM (L): OCT 14: -3.40% to $430.00 (cum +1.59%); broke 50d pre-print. Q3 TONIGHT Oct 15 AMC (framework #4 already applied 4.0->3.5 Oct 12). Hold 3.5%. | *3nm/2nm ramp; CoWoS expansion; AI chip demand*
• DELL (L): OCT 14: -11.27% to $507.93 (cum +14.66%); Barclays downgrade to Equalweight ('enterprise AI timing', only ~25% of enterprise AI spend to hardware) + 50d break -> TRIMMED 3.0->2.0 (fresh opposing marquee-analyst catalyst, CRM Oct 13 family; mechanism claim attacks backlog-conversion timing). Nov 24 print. | *AI server backlog $95B; record $60.9B orders; enterprise refresh*
• ANET (L): OCT 14: -4.45% to $195.42 (cum +11.15%); broke 50d (76d above) on optical/AI-funding rout. Nov 3 print. Hold 3.0% (banked Oct 7). | *AI back-end networking; 800G/1.6T switch cycle*
• MSFT (L): OCT 14: -1.80% to $533.69 (cum +10.92%); mega-cap de-risk, held better than complex. Oct 28 print. Hold 3.5%. | *Azure/AI cloud demand; Copilot monetization; Cantor PT $608*
• AMD (L): OCT 14: -5.73% to $570.82 (cum +21.59%); high-beta de-risk day 2. Oct 29 AI DevDay + Nov 3 print. Hold 2.5% (double-banked). | *MI400 series ramp; server share gains vs Intel*
• ADI (L): OCT 14: -2.68% to $403.57 (cum +4.76%); analog bid reverse. Nov 24 print. Hold 4.0%. | *Above-seasonal Q1 guide; 10/11 qtrs bookings growth; DD growth 2027; broad-market recovery*
• PANW (L): OCT 14: -5.54% to $375.04 (cum -4.64%); security-complex de-risk, spend narrative intact. Nov 18 Q1 FY27 print. Hold 3.5%. | *AI-safety debate driving security spend; platform consolidation; analyst AI-targets*
• CRWD (L): OCT 14: -4.56% to $249.29 (cum -5.03% = band zone); software de-risk. Dec 1 print. Hold 3.0%. | *AI-driven cyberthreat narrative; Wall Street AI-security PTs; platform consolidation*
• NXPI (S): OCT 14: stock -2.43% to $220.83 (cum +20.36% on CORRECTED simple basis; the Oct 13 trim fired on a covering-basis inflation at +23.12% vs correct +18.78% - RE-UNDERWRITTEN: trim stands (realized $225.21 vs $220.83 now) and framework #3's +20% trigger is only NOW genuinely met; second trim deferred to the framework #4 pre-print trim into Oct 27). Hold 2.5%. | *Automotive semi demand weakness; China EV slowdown*
• AMZN (S): OCT 14: stock -1.74% to $254.08 (cum +4.94%); back below 50d (5d above). Oct 29 print. Hold 3.5%. | *AWS growth deceleration; retail margin compression; regulatory overhang*
• ARM (S): OCT 14: stock -6.81% to $238.01 (cum +26.29% CORRECTED basis) -> TRIMMED 2.5->2.0 (framework #3, first bank since Sep 21 entry; GFS multi-trim precedent). Nov 4 print = framework #4 trim next. | *CPU rotation extension; valuation mean reversion*
• SNDK (S): OCT 14: stock -5.33% to $1,403.34 (cum +25.63% corrected); CXMT supply unwind paying. Oct 29 print. Hold 2.5% (banked Oct 7). | *Barron's cover hype; NAND supercycle peak; mean reversion*
• CRM (S): OCT 14: stock -0.87% to $235.95 (cum +0.69% corrected); below the $242 close-trigger. Dec 2 print. Hold 2.5%. | *Agentic AI disruption to seat-based SaaS; application-layer commoditization*
• UMC (S): OCT 14: stock -4.42% to $21.40 (cum +13.29% corrected); broke 50d (36d above), mature-node weakness intact. Oct 28 print. Hold 1.5%. | *Mature-node weakness; no AI exposure; semi selloff*
• WDC (S): OCT 14: stock -4.36% to $364.10 (cum +23.14% corrected); memory unwind paying. Oct 22 print. Hold 1.5% (banked Oct 7). | *Memory/storage profit-taking; positioning unwind; June peak mean-reversion*
• ADBE (S): OCT 14: stock +1.86% to $240.68 (cum -2.21% corrected); value-buyer bounce off the 63% YTD drawdown = positioning, no mechanism change. Dec 9 print. Hold 3.5%. | *AI disruption to creative SaaS; soft forward guidance; SaaS multiple compression*
• QCOM (S): OCT 14: stock -2.40% to $172.62 (cum +7.93% corrected); broke 50d (36d above), BOM-squeeze intact. Nov 4 print. Hold 4.0%. | *JPM downgrade PT $140; smartphone shipment decline; memory-cost margin squeeze*
• NOW (S): OCT 14: stock +0.99% to $139.98 (cum -4.17% corrected); Citi PT-revamp bid. Oct 28 print = event risk. Hold 2.5% (watch). | *Anthropic Managed Agents disruption; seat/workflow-automation commoditization; Oct 28 earnings reckoning*
• NFLX (S): OCT 14: stock +0.16% to $70.57 (cum -5.23% corrected, watch); WF PT $57 thesis intact into the Oct 20 binary (2 sessions). Hold 2.0%. | *WF downgrade PT $57; streaming subscriber growth concerns; Oct 20 earnings*
• GLW (S): OCT 14: stock -5.15% to $140.89 (cum +11.60% corrected); paid on the optical rout BUT trailed peers again (AAOI -11.82%/CIEN -10.19%) = relative-weakness confirmation leg still absent. Oct 27 print. Hold 1.0% core. | *$2B ATM equity offering (dilution overhang); relative weakness vs optical peers; Oct 27 earnings*
• CAT (S): OCT 14: stock -3.08% to $748.98 (cum +7.97% corrected); FTC/USDA overhang grinding. Oct 28 print. Hold 2.5%. | *FTC/USDA farm-equipment competition inquiry (regulatory overhang); 50d break; soft construction/China cycle*
• ASMIY (S): OCT 14: stock -2.94% to $904.16 (cum +10.50% corrected); ASML beat/raise + '2027/28 output will not fully meet demand' wobbles the ALD-capex-normalization leg BUT the WFE tape de-rated regardless (ASML -3.94% sell-the-news) = guidance-miss + complex-de-risk legs intact. Oct 27/28 print. Hold 2.5%. | *FY revenue guidance below Street (WSJ); WFE complex de-risking; ALD capex-cycle normalization*
• IBM (L): OCT 14: -1.26% to $227.88 (cum +0.49%); defensive quality holding in the rout. Oct 21 Q3 print (framework #4 pre-print trim call on the Oct 19-20 runs). Hold 3.5%. | *Analyst upgrade to Buy PT $260; quantum computing progress; VMware services handoff; Oct 21 Q3 print*
• SKHY (S): OCT 14: stock -6.06% to $147.55 (cum +9.05% corrected); CXMT memory unwind paying. Oct 27 print. Hold 2.0%. | *CXMT DRAM-capacity progress (fresh supply-side datapoint); memory supercycle peak; Toshiba-HDD-glut complex overhang; Oct 27 print*
• TSEM (L): OCT 14: -5.87% to $209.16 (cum -5.87%, watch); optical/foundry contagion day 2 (AAOI -11.82%/CIEN -10.19%) despite the Semtech SiGe deal leg intact. Nov 2 Q3. Hold 2.5%. | *Semtech SiGe capacity deal; AI data-center silicon-photonics/SiGe interconnect demand; NVIDIA 1.6T optical; Nov 2 Q3*
• AMAT (L): OCT 15: RE-ENTRY long 2.5% at $446.63 (Oct 14 close, cum 0.00%) - the Oct 14 risk-close reversed on a NEW distinct catalyst (QCOM Sep 29 re-entry family, close reason explicitly re-underwritten): ASML Q3 beat+RAISED - planned 2027/2028 output 'will not fully meet demand', customers asking for 2029/2030 POs, value-based pricing COMPLETED (full effect 2028), FY26 raised to EUR44-45B = the WFE demand backdrop the AI-capex/WFE-leader thesis needed. The binary that forced the Oct 14 close RESOLVED BENIGNLY and the EPIC gripe (no 2030 target) is a nit vs an industry sold out through 2028-30. Entered after -3.79% (464.24->446.63) with the tape ignoring the print (CPI + CRWV debt panic) = no gap-chase. NOT re-paired with ASMIY (pair failed on the long leg Oct 13; ASMIY short kept standalone). Nov 12 Q3 (est) = defined event risk. | *ASML demand sold out 2027/28 + 2029/2030 PO requests + value-based pricing; AI-capex WFE leadership; EPIC Center ramp; Nov 12 Q3*
• AAPL (L): OCT 15: OPENED long 2.5% at $342.96 (cum 0.00%) - fresh analyst + company catalysts: BNP Paribas upgrade to Outperform + HomeView smart-home initiative (Oct 13-14 home-products event) + defensive-bid relative strength (+0.50% in a -1% tech rout while DELL -11.27%). Book-level offset to the structural QCOM short (iPhone-cycle proxy vs handset BOM squeeze). Nov 2 print = defined event risk. Sized 2.5% (analyst+initiative tier, IBM precedent). | *BNP Paribas upgrade to Outperform; HomeView smart-home initiative; iPhone Duo cycle; Nov 2 Q3*
• ORCL (S): OCT 15: OPENED short 2.0% at $130.55 (cum 0.00%) - AI-infrastructure FINANCING-COST mechanism (cross-sector narrative test, CRM Sep 23 family): CRWV bonds trading ~15% yield (9.625% coupon) = fresh Oct 14 sector cost-of-capital shock for debt-funded AI capex; ORCL is the purest watchlist expression (negative FCF, FY27 capex >=$90B above est, $664B RPO needing debt-funded buildout - the exact 'capex above estimates + negative FCF' legs flagged Sep 10). Tape-confirmed: -8.83% vs Big Tech -0.5/-1.8% = 5x underperformance on the same theme. Pairs AAPL 2.5/MSFT 3.5 longs vs AMZN 3.5/ORCL 2.0 shorts = Big Tech L6.0/S5.5 (+0.5). PRE-COMMITTED EXIT: financing narratives whipsawed before (Sep 10->11 flip on demand data) - trim/close on an AI-capex melt-up squeeze, TSM print tonight = the squeeze trigger to watch. Dec 9/14 print. Sized 2.0 (financing/capital-markets tier, GLW precedent). | *CRWV 15% bond yield = sector financing-cost shock; negative FCF + capex-above-est; RPO-to-revenue conversion risk; Dec 9/14 print*
• SPCX (S): OCT 15: OPENED short 2.0% at $152.16 (cum 0.00%) - neocloud financing-cost family, contained-chase expression of the Oct 14 CRWV debt panic (CRWV -15.46%/NBIS -10.08% evaluated and REJECTED as single-day chases; SPCX only -2.80% after its own Oct 13 -4.21% spectrum/debt concerns = own-balance-sheet leg + sector cost-of-capital shock). Neoclouds fund AI capex with high-cost debt; CRWV's ~15% bond yield re-prices the sector's funding costs. Nov 3 print (est) = defined event risk. Sized 2.0 BELOW the 2.5 norm on chase risk (SKHY Oct 13 precedent). | *CRWV debt-panic sector cost-of-capital repricing; own spectrum/debt concerns; AI-capex financing squeeze; Nov 3 print*

📋 Portfolio Rules