Overview

本周（2026-08-31 至 09-06，UTC）Trading log 的主线是：半导体配置保持核心地位，但操作从前段增加风险，转向后段削减杠杆、部分止盈。最新完整组合数据仍是截至 9 月 3 日的 Flex 日终快照；其后的 SOXL、加密资产减仓为作者自述，不能直接折算为最新组合权重。周末作者补充了加密市场参与定价的半导体参考价，同时明确其不代表美股正式开盘价格。

Key Themes

• 杠杆先增后减：[Post 3] 报告 TQQQ 加仓至 3.96%，战术杠杆升至 13.06%；[Post 9] 随后确认 TQQQ 减至 1.96%，合计杠杆降至 10.68%。再后的 [Post 12] 自述卖出三分之一 SOXL，因此 10.68% 不是该笔操作后的比例。
• 流动性修复，但未达到日报目标：一级流动缓冲从 [Post 1] 的 6.40% 降至 [Post 3] 的 4.39%，最新快照回升至 9.48%；[Post 9] 仍提出新增风险前恢复至 10% 以上。
• 集中度看共同因子，而非产品数量：日报持续将 DRAM、DISK 归为同一内存主题 ETF 袖套；[Post 9] 核验后的半导体／存储总风险链条为 48.31%。各日报统计范围有所调整，不宜把所有宽口径数字直接当作同口径增减。
• 长期方向与战术降压并存：[Post 5] 表示继续布局半导体设备、先进封装、光互连，并计划按 4:1—3:1 定投 SMH 与 NVDA；后续减持杠杆并未明确撤回这一方向。日报则持续主张限制重复敞口，二者不能合并解读为全面加仓。
• 收益分配不被视为现金安全垫：GPIQ／OVL 主袖持续接近均衡，最新合计 34.29%；日报反复强调其与 CHPY、MSTY 仍承担权益或相关标的风险，高分配不等于低风险。[Post 9]

Notable Posts

• [Post 3] → [Post 9] → [Post 12]：从加仓 TQQQ 转为减仓 TQQQ，随后作者称在 SOXX 触及 520 时卖出三分之一 SOXL，构成本周风险回收的主要变化。
• [Post 7] → [Post 8] → [Post 10]：BTC 表述从关注“二探”后的突破，转为“继续出货剩下的 1/7、利润转去美股”，随后强调追高风险与不追高偏好；不能仅保留早先的偏多观察。
• [Post 5]：明确半导体设备、先进封装、光互连的持续配置意向，以及 SMH／NVDA 的计划比例；帖子未确认定投已经执行。
• [Post 13]：作者称在 BNB 上涨时卖出八分之一现货，延续周后段部分落袋的操作。
• [Post 14]：作者称加密市场参与定价的半导体处于“多头状态”，给出 SOXX 约 530、SOXL 约 124、SMH 约 575；同时提示次日美股休市，提前定价不能代表正式开盘价。

Watchlist

• SOXL 减仓后的实际组合权重、总杠杆及现金缓冲，尚无后续 Flex 快照确认。[Post 9][Post 12]
• DRAM／DISK 持仓重叠与共同周期暴露仍待复核；日报提出的流动缓冲超过 10% 目标，在最新快照中尚未达到。[Post 9]
• SMH／NVDA 定投及半导体细分配置计划、加密利润转入美股的后续执行情况，尚未见完整确认。[Post 5][Post 8]

Overview

本周交易主线集中在半导体、收益型权益与加密相关仓位，期间多次调整杠杆和流动性。组合先提高 SOXL、DRAM、DISK 等高波动敞口，随后减持 SOXL、清仓 MSTR 与 TENX；截至最新快照，一级流动缓冲升至 25.24%，但半导体/存储相关性风险仍较集中。

Key Themes

• 风险先升后降： 8月24日快照显示流动缓冲从 15.62% 降至 7.12%，半导体/内存袖套升至 43.49% [Post 8]；之后减持 SOXL，缓冲回升至 11.46% [Post 9]，最新进一步升至 25.24% [Post 18]。
• 半导体仍是核心风险簇： 最新确认的直接半导体/存储袖套为 34.97%，CHPY、SMH、SOXL、AMKR、DRAM、DISK 存在重叠，SOXL 还附带每日杠杆和路径风险 [Post 18]。
• NVDA 交易由观察转为快进快出： 先关注 205 位置及财报 [Post 2][Post 11]，随后小额买入并在约 218 卖出；作者明确表示不考虑单独长期持有 NVDA，继续偏向半导体 ETF [Post 12][Post 13]。
• 加密敞口主动收缩： 作者称 BTC 在两个局部高位累计卖出三分之一 [Post 5]，之后清仓 MSTR [Post 18]；但 MSTY 仍占 2.19%，因此相关风险并未完全退出。
• 收益主袖趋于均衡： OVL 获加仓确认 [Post 16]；最新 GPIQ 与 OVL 分别为 17.44% 和 17.48%，接近 50/50，但两者仍属于权益及期权策略风险，不是现金替代 [Post 18]。

Notable Posts

• [Post 8][Post 9] SOXL、DRAM、DISK 加仓一度显著推高风险，随后 SOXL 从 11.52% 减至 7.81%，体现从扩张杠杆转向控制尾部风险。
• [Post 12][Post 13] NVDA 在 205 附近小买，财报后约 218 卖出，被明确界定为短线交易；此前等待财报再做更大决定的计划由此结束。
• [Post 18][Post 19] 最新快照确认清仓 MSTR 和 TENX；作者随后表示 MSTR 主要参照 BTC 走势交易，并认为离场及时。
• [Post 20] 作者称近一个月短线期货交易约贡献当月收益的 25%；所附 AI 汇总显示严格口径为 6 笔、胜率 83.3%、净收益约 487 美元，但样本较小且缺少实际保证金与杠杆数据。

Watchlist

• 半导体/存储直接袖套仍为 34.97%，且与 GPIQ、TQQQ 的科技敞口存在进一步重叠 [Post 18]。
• SOXL 与 TQQQ 合计 8.85%，杠杆及路径风险仍在 [Post 18]。
• MSTR 已清仓，但 MSTY 2.19% 继续保留 MSTR/BTC 相关敞口 [Post 18]。
• SMH 正测试 561.72–568.51 压力带，帖子尚未确认有效突破 [Post 18]。
• BTC 是否突破作者标注的位置仍未解决；其当时判断是未突破则维持“lower high、反弹非反转” [Post 17]。

Overview

本周组合明显转向进攻：QQQI 被清仓并新建 OVL，SOXL、DRAM、DISK 与 NBIS 获得配置，一级流动性由约 23% 降至约 11%。半导体/存储与杠杆风险成为核心议题，同时频道记录了多笔加密货币止盈及继续持有现货的安排。所有组合日报均为 IBKR T-1/EOD 快照，并非盘中实时持仓。

Key Themes

• 最新日报估算半导体/存储链条占比约 38.93%，SOXL 与 TQQQ 战术杠杆仓合计约 11.05%，两者均接近或超过一级流动性规模。
• 收益主袖由 GPIQ 与 OVL 构成，合计约四成且接近 50/50；帖子强调这些覆盖策略仍属于权益风险，并非现金替代。
• SOXL 连续加仓，DRAM 与 DISK 被视为同一内存周期风险组；NBIS 则作为 AI 基建观察仓新建。
• 加密交易方面，发帖者报告 BTC 合约止盈、BNB 合约继续持有后退出，以及在接近 8 万美元时卖出四分之一 BTC 现货。
• 个股调整包括卖出约 60% MSFT 仓位、清仓 DUOL 与 OKB；MSTR、MSTY 及剩余 BTC 现货仍计划继续持有。

Notable Posts

• [Post 2] 报告清仓 QQQI 并新建约 19.94% 的 OVL，收益主袖由此转向 OVL 与 GPIQ。
• [Post 7] 记录 SOXL 扩至 6.81%，DRAM 与 DISK 合计约 7.82%，一级流动缓冲降至 12.11%。
• [Post 25] 显示 SOXL 进一步升至 8.53%、DUOL 清仓、NBIS 新建至 2.37%，并提示 AI/半导体相关性继续上升。
• [Post 16] 发帖者称 BTC 4倍合约从 64,659.1 开仓、于 68,666 平仓；随后又表示已退出币本位合约并保留现货。[Post 31]
• [Post 26] 将个体化 mRNA 黑色素瘤疗法描述为重要临床进展，同时注明完整Ⅲ期数据、总生存期、监管审批及其他癌种验证仍未完成。

Watchlist

• CHPY、SMH、SOXL、DRAM、DISK 与 AMKR 的相关风险，以及约四成半导体/存储集中度。[Post 35]
• NBIS 观察仓的合同转化、资本效率、融资与稀释情况。[Post 25]
• MSTR 与 MSTY 的继续持有及“翻倍后取回本金”计划。[Post 36]
• MSFT 剩余仓位及下月底 ESPP 到账后的处理。[Post 3]
• 个体化 mRNA 黑色素瘤疗法的完整Ⅲ期结果、总生存期随访与监管进展。[Post 26]

Overview

• The supplied posts in this window were concentrated on Aug 10–14 and were mostly daily portfolio risk reports based on IBKR Flex LastBusinessDay data, not intraday snapshots.
• The portfolio stayed centered on GPIQ + QQQI, which moved from 42.23% (Aug 10 report) to 38.88% (latest supplied report on Aug 14), while remaining near a 51/49 internal split.
• By Aug 14, positioning had become more defensive: cash/liquidity buffer rose from 17.10% to 25.93%, AI/semiconductor exposure fell from 31.01% to 25.61%, and TQQQ + SOXL dropped from 6.48% to 5.13%.
• CHPY stayed above 11% throughout the week; SMH declined from 13.12% to 10.09% after trimming.
• Reported holding count increased from 11 to 13.

Key Themes

• Rebalancing / de-risking: Reported changes included a SOXL trim, IONQ exit, new small positions in MSTY and TENX, and later reductions in SMH, SOXL, DUOL, and FISV.
• Role-based portfolio framing: Posts repeatedly separated the book into Nasdaq income core (GPIQ/QQQI), semiconductor core/satellite (SMH/CHPY), and tactical leverage (TQQQ/SOXL).
• Resistance/pressure testing: Multiple reports flagged GPIQ, QQQI, SMH, and CHPY as testing near-term pressure/resistance zones.
• AI infrastructure focus: The Nebius analysis argued AI demand remained strong, but the CapEx → profit/FCF loop was not fully closed, given reliance on prepayments, debt, and equity financing.
• Research ideas/tools: A long-term ETF return site was recommended, and DRAM / DISK were mentioned as memory ETF ideas to watch.

Notable Posts

• Aug 10 risk report: SOXL cut from 5.05% to 3.68%; IONQ fully exited (1.52% to 0%).
• Microsoft ESPP note: The author said the entire Microsoft ESPP position was sold at the open, with profit described as satisfactory.
• Aug 11 risk report: New MSTY position at 2.29%.
• Aug 12 risk report: New TENX position at 0.59%.
• Aug 12 AI CapEx post (Nebius): Reported AI cloud revenue +514% YoY, ARR run rate of $3.0B, customer commitments above $40B, but rough Q2 FCF around -$3.41B.
• Aug 13 note: Short post saying DUOL was performing well that day.
• Aug 14 risk report: Largest rebalance of the week: SMH cut to 10.09%, SOXL to 2.49%, DUOL to 1.48%, FISV to 0.95%; cash buffer up to 25.93%.
• Tool recommendation: totalrealreturns.com was recommended for checking long-term ETF returns.

Watchlist

• GPIQ / QQQI: Still the main exposure block; posts repeatedly said future adds should depend on QQQ support and available cash rather than mechanical rebalancing.
• CHPY: Stayed a double-digit semiconductor income satellite all week and was repeatedly described as large enough to materially affect drawdowns.
• SMH: Remained a core semiconductor holding after trimming; several reports flagged it near resistance.
• TQQQ / SOXL: Reduced during the week but still treated as tactical risk capital, not long-term core exposure.
• DRAM / DISK: Explicitly mentioned on Aug 12 as ETF ideas to monitor if Nasdaq leads and semis follow.
• Nebius follow-through: The AI CapEx post highlighted ongoing checks around customer commitment quality, prepayment terms, GAAP profitability, utilization, pricing, and free cash flow.

Overview

• Posts supplied for this window were concentrated on Aug 3–5.
• The week’s tone was more defensive than aggressive: several posts favored waiting, trimming, or not adding rather than chasing moves.
• Portfolio snapshots came from IBKR Flex `LastBusinessDay` reports, so they were not intraday real-time holdings.
• Reported concentration stayed high:
  • Aug 3 report: JEPQ 35.58%, SMH 14.15%, CHPY 11.78%; primary liquidity buffer 19.17%.
  • Aug 4 report: GPIQ 21.85%, QQQI 21.09%, SMH 12.82%, CHPY 10.65%; liquidity buffer 15.40%.
  • Aug 5 report: GPIQ 21.60%, QQQI 20.81% (combined 42.41%), SMH 13.03%, CHPY 10.91%; liquidity buffer 17.56%.
• The Aug 5 report explicitly identified one real rebalance: TQQQ reduced from 5.32% to 2.82%.

Key Themes

• Risk control first. The channel repeatedly stressed position discipline, pressure/resistance awareness, and reducing risk if needed.
• Income ETFs remained core exposure. JEPQ early in the week, then GPIQ + QQQI became the reported center of the portfolio. The posts also stressed that yield vehicles are still equity risk, not cash.
• Semiconductors stayed important, mainly through SMH, CHPY, SOXL, AMKR, with repeated warnings against stacking overlapping exposure or adding leverage casually.
• Leveraged trades were treated as tactical/speculative, not core holdings. This was stated directly for TQQQ/SOXL and echoed in the posts on RKLB, ORCL, and BNB perpetuals.
• Radar / resistance zones were a recurring decision tool, especially for QQQI, SMH, CHPY, plus screenshot alerts for COHR, MU, NBIS, KLAC, ASML, AMAT.

Notable Posts

• Aug 3: Said they did not add SOXL; noted half had been sold the prior week at 128, with remaining cost basis 95.09.
• Aug 3: For TQQQ, said cost basis was 59.91 and planned to start taking profits around 75, roughly corresponding to QQQ near 715.
• Aug 3: On RKLB, said it was worth watching but preferred spot shares over leverage; also shared a small 5x perpetual trade screenshot.
• Aug 3: On ORCL, said they closed the position rather than betting on a breakout; shared a +38.44% perpetual trade screenshot.
• Aug 4: Posted a BNB 5x long screenshot and described it as a discretionary “bet.”
• Aug 4: Said NBIS holders could consider taking profits, noting it was trading largely independently from CRVW.
• Aug 4: Posted “I reduced” with a QQQ chart.
• Aug 4: On CHPY, warned that its high distribution can tempt investors to go all-in, but drawdowns still behave like equity; said they did not plan to add unless there was a black-swan-type selloff.
• Aug 5: Said the day was not suitable for trading, and if anything had to be done, it should be reducing positions.

Watchlist

• GPIQ / QQQI — dominant reported weights by Aug 4–5; QQQI was flagged at first resistance on both 4H and 1D.
• SMH — core semiconductor ETF; listed as testing 1D first resistance 554.98–567.72 in the Aug 4 report.
• CHPY — two-digit portfolio weight; Aug 5 report flagged 4H first resistance 71.33–72.47 and 1D first resistance 70.61–72.24.
• TQQQ / SOXL — tactical leveraged exposure; TQQQ was trimmed in the Aug 5 report, while SOXL was monitored but not added.
• NBIS — explicitly mentioned as a possible take-profit candidate.
• Radar names shown in screenshots: COHR, MU, KLAC, ASML, AMAT.

Overview

Based on the supplied posts in this window, activity centered on two public portfolio risk reports and several tactical notes on semis, Nasdaq, and BTC.

| Metric | 2026-07-27 post | 2026-07-28 post | |---|---:|---:| | Primary liquidity buffer (cash + SGOV/BIL) | 26.78% | 11.66% | | SMH weight | 27.06% | 29.39% | | JEPQ weight | 20.96% | 28.13% | | AI / semiconductor-chain exposure | 40.78% | 43.91% | | Semicap equipment exposure | 7.76% | 7.55% | | Holdings count | 9 | 10 |

• Both daily reports used IBKR Flex `LastBusinessDay` data, not intraday real-time holdings.
• Portfolio concentration remained high in SMH and JEPQ.
• Between the two reports, JEPQ increased by 7.17 pct and the primary liquidity buffer fell by 15.12 pct.
• Outside the reports, posts focused on semiconductor weakness, support-based adds, and BTC support / stop-loss planning.

Key Themes

• Concentration and liquidity management
  • SMH and JEPQ were repeatedly identified as the main drivers of portfolio volatility.
  • JEPQ was explicitly treated as income equity / second-layer liquidity, not cash.

• Heavy semi/AI exposure
  • Posts tied portfolio risk to overlapping exposure in SMH + AMAT and broader AI/semiconductor names.
  • An ASML-linked news item and charts of ASML, AMAT, AMD, INTC, AMKR, MU were used to frame current weakness.

• Adding on weakness, but near support
  • The reports stressed adding only on clear support / pullback confirmation for oversized positions.
  • Separate posts showed a willingness to add selectively: e.g. AMKR was bought back after having been sold above 70 earlier.

• Differentiation inside storage / memory
  • A specific note said SNDK is more cyclical, and should not be treated the same as MU / SKHY.

• Crypto risk management
  • BTC was described as repeatedly testing support; if support broke, the plan was to stop out the remaining spot position and reallocate capital elsewhere.
  • A later BTC post said a turning point had appeared and was roughly in sync with Coinbase.

Notable Posts

• 2026-07-27 — Public portfolio risk report
  • Highlighted concentration in SMH 27.06% and JEPQ 20.96%.
  • Reported AI/semi exposure at 40.78% and primary liquidity buffer at 26.78%.
  • Mentioned tested zones for SMH, JEPQ, AMAT, SKHY, QQQM.

• 2026-07-27 — Nasdaq / semi weakness thread
  • A weekly QQQM/Nasdaq chart marked a lower area as a potential leverage-entry zone if price fell there.
  • An ASML news link and chart were paired with AMAT as a reference for how semi names were reacting.
  • A separate note differentiated SNDK from MU / SKHY on cyclicality.
  • The author also posted a reflection on investing/trading as a process of improving knowledge.

• 2026-07-28 — Public portfolio risk report
  • Showed SMH 29.39%, JEPQ 28.13%, and AI/semi exposure 43.91%.
  • Noted JEPQ +7.17 pct vs. the prior report and primary liquidity buffer down to 11.66%.
  • Flagged JEPQ 57.34–57.63 as a 4H support zone being tested.

• 2026-07-28 — BTC / semis positioning
  • BTC was described as probing support; a break would trigger a stop on the remaining spot.
  • Another BTC post said a turning point had appeared and matched Coinbase behavior.
  • AMKR was described as hitting weekly support, with some shares bought back.
  • A later post said the account was fully allocated, with indices making up about 70% combined; MU would not be added because no more cash would be deployed.

Watchlist

• SMH / AMAT / ASML
  • Main semi risk cluster in the posts.
  • Overlap between SMH and AMAT remained a recurring concern.

• JEPQ
  • Large portfolio weight.
  • Reported levels: 57.34–57.63 (4H support) on Jul 28; 58.47–58.79 (1D support) and 58.64–59.03 (4H resistance) on Jul 27.

• QQQM / Nasdaq
  • Weekly chart highlighted a deeper downside zone as a possible leverage area.
  • Jul 27 report also cited 285.71–287.68 as a 4H resistance zone.

• BTC / Coinbase
  • Support integrity remained the key trigger in the posts.
  • A break was tied to reducing spot exposure.

• AMKR / MU / SNDK / SKHY / AMD / INTC
  • Used as read-through names for semi and storage weakness.
  • SNDK was explicitly flagged as more cyclical than MU / SKHY.

Overview

• Three public portfolio risk reports were posted on Jul 20, Jul 21, and Jul 22. Each noted that the IBKR Flex data were last-business-day snapshots, not intraday real-time holdings.
• Across those reports, SMH and JEPQ were repeatedly identified as the main portfolio risk drivers. Reported cash / cash-like buffer ranged from 16.82% to 46.58%.
• The reported holding count fell from 15 (Jul 20) to 13 (Jul 21) to 9 (Jul 22).
• Outside the risk reports, posts focused on BTC trading structure, rotation out of lower-utilization positions, and semiconductor / tech setups such as AMAT, SKHY, NBIS, and GOOG.

Key Themes

• Concentration and exposure control:

SMH and JEPQ stayed large in the portfolio. AI / semiconductor-chain exposure was reported at 50.74% (Jul 20), 51.15% (Jul 21), and 35.78% in the corrected-weight Jul 22 report.

• Capital rotation:

A post said SPCX and ORCL were cut because capital utilization was too low and opportunity cost was too high. Another post said capital had been moved into SKHY.

• BTC as a trading, not passive, asset:

Posts discussed Binance liquidations, exchange-specific wick behavior versus Coinbase, a partial BTC sale, a possible short-term consolidation/topfinder setup, and interest in buying some BTC back around 63,000.

• Semi/tech ideas remained active:

AMAT and SKHY were highlighted positively; NBIS was described as showing rejection; GOOG was named as a buy candidate if it reached 270.

Notable Posts

• Jul 20 risk report:

Reported SMH 40.17%, JEPQ 14.08%, QQQ 8.37%, AI/semi exposure 50.74%, and cash 16.82%. It also noted QQQ increased by 8.36 pct and cash had declined from 34.36% to 16.82%.

• Jul 20 BTC post:

Said Binance alone had liquidated about $140M of BTC longs and that Coinbase did not show the same sharp drop seen on Binance.

• Jul 21 risk report + trade note:

Reported SMH 46.37%, JEPQ 20.26%, AI/semi exposure 51.15%, and cash 29.92%. A separate post said SPCX and ORCL were sold, and some BTC was also sold.

• Jul 21 discretionary posts:

A post said the move into SKHY looked like a good choice. Another said AMAT remained trustworthy. A screenshot showed MU +12.58%, SNDK +12.66%, and SKHY +12.74%.

• Jul 22 corrected-weight risk report:

Reported SMH 27.52%, JEPQ 14.05%, AI/semi exposure 35.78%, cash 46.58%, and 9 holdings.

• Jul 22–23 setup posts:

NBIS was flagged as “seeing rejection”; GOOG was noted as interesting at 270; BTC was marked as a possible buyback around 63,000.

Watchlist

• SMH / JEPQ: still the clearest portfolio concentration points. JEPQ pressure zones mentioned were 59.17–59.49 (Jul 21) and 59.02–59.31 (Jul 22).
• BTC: watch whether price revisits ~63,000; prior posts also flagged exchange-specific volatility and value-area/LVN acceptance.
• GOOG: explicit interest if it drops to 270.
• NBIS: rejection was noted, so follow-through matters.
• SKHY / AMAT: both were highlighted positively in discretionary commentary.

Overview

• The week’s supplied posts focused mainly on semiconductor/AI names, plus a few Binance perpetual screenshots.
• Early setup posts highlighted AMAT (“local bottom starting to show up”), MU (a 5x long), and SNDK (“worth trading/speculating”).
• Two public portfolio risk reports — both based on IBKR Flex `LastBusinessDay` data, not intraday real-time holdings — showed a concentrated book in SMH, JEPQ, AMAT, and ORCL.
• From the 2026-07-13 report to the 2026-07-14 report, AI/semiconductor exposure increased from 49.86% to 53.55%, while cash / cash-like buffer fell from 48.69% to 37.61%.
• Later posts showed positive 5x perpetual screenshots in AVGO, ORCL, NOW, MU, and AMAT, plus a note about a missed ETH pattern and profit-taking ahead of ASML earnings.

Key Themes

• Semiconductor concentration: AMAT, SMH, MU, SNDK, and ASML were recurring names across charts, risk reports, and trade screenshots.
• Risk discipline over chasing: Both portfolio reports stressed managing entries/exits around support/resistance rather than blindly adding to already large AI/semi exposure.
• AMAT became more important: It was flagged as a possible local bottom, and its portfolio weight rose 11.16% → 17.90% between the two risk reports.
• Tactical leveraged longs: Screenshoted perpetual positions showed positive P/L at capture time: AVGO +11.32%, ORCL +8.17%, NOW +21.98%, MU +15.83%, AMAT +14.76%.

Notable Posts

• 2026-07-13 Portfolio risk report
  • Top weights: SMH 28.30%, JEPQ 18.50%, ORCL 13.03%, AMAT 11.16%
  • AI/semi exposure: 49.86%
  • Cash / cash-like buffer: 48.69%
  • Holdings count: 14

• AMAT / MU / SNDK setup posts
  • AMAT: “local bottom starting to show up”
  • MU: “open a long here, 5x”
  • SNDK: described as a tradable/speculative spot, followed by a later chart update

• 2026-07-14 Portfolio risk report
  • Top weights: SMH 27.32%, JEPQ 18.54%, AMAT 17.90%, ORCL 10.23%
  • AI/semi exposure: 53.55%
  • Cash / cash-like buffer: 37.61%
  • AMAT weight change: 11.16% → 17.90%
  • Holdings count: 15

• 2026-07-14 to 2026-07-15 perpetual screenshots
  • AVGOUSDT 5x long: +11.32%
  • ORCLUSDT 5x long: +8.17%
  • NOWUSDT 5x long: +21.98%
  • MUUSDT 5x long: +15.83%
  • AMATUSDT 5x long: +14.76%

• ETH / ASML notes
  • One post said an ETH pattern was missed due to focus on U.S. equities, and was logged without hindsight claims.
  • Another post said ASML profits were taken ahead of its earnings day.

Watchlist

• AMAT — repeated across setup posts, portfolio concentration, and perpetual screenshots.
• SMH / JEPQ / ORCL — remained the main portfolio drivers in the risk reports.
• MU / SNDK — both were highlighted as tactical chart setups during the week.
• ASML — earnings-related name after pre-earnings profit-taking.
• ETH — pattern noted in hindsight-free post, but not traded in the supplied material.

Overview

• The supplied posts for the week were concentrated on Jul 6–8 and focused on semiconductor-heavy portfolio risk, adding into weakness in select names, and exiting BTC exposure.
• Public portfolio snapshots stayed semi/AI-heavy:
  • Jul 6 report: SMH 39.41%, AMAT 15.05%, ASML 14.72%; AI/semis 84.83%; cash/class-cash 41.16%.
  • Jul 7 report: SMH 37.39%, AMAT 15.72%, ORCL 10.16%, KLAC 8.25%; AI/semis 74.39%; cash/class-cash 45.17%.
• Both risk reports explicitly note they are IBKR Flex `LastBusinessDay` snapshots, not real-time intraday holdings.

Key Themes

• Semis remained the main book driver: SMH and AMAT were repeatedly highlighted as the biggest sources of portfolio volatility.
• Selective dip-buying: the selloff was described as not systemic risk, and the stance was to keep holding while adding back to names like AMAT.
• ORCL exposure increased: a leveraged ORCL position was opened alongside added stock exposure; the next risk snapshot showed ORCL up from 4.67% to 10.16%.
• BTC exposure was reduced/closed: BTC was monitored closely, then a post said the position was closed and spot BTC was sold.
• Single-name selectivity: GOOG was treated cautiously, while MU was watched for a support/resistance-based long setup.

Notable Posts

• Jul 6 — ORCL: “Opened leverage and added stock.” The attached Binance perpetual screenshot showed ORCLUSDT 5x long at +12.14%.
• Jul 6 — BTC: BTC was posted as “watching 👀”, then later “I closed BTC” and spot BTC was sold around 64,047.56.
• Jul 7 — BTC flow comment: the post said heavy sells around 60,000 were absorbed, describing them as mostly stop orders.
• Jul 7 — MU: watched a posted S/R zone; if price was rejected lower and held, the plan was to buy calls/long with a stop under 1%.
• Jul 7 — portfolio shift: the next risk report showed ORCL +5.49 pct and KLAC +4.33 pct versus the prior snapshot.
• Jul 7–8 — AMAT: the author said they were slowly adding AMAT back, after previously trimming around 722, and later added that other positions were also increased.

Watchlist

• SMH / AMAT: still the main portfolio volatility drivers; the posts repeatedly emphasized support/resistance discipline over chasing.
• ORCL: active trade focus and now a larger portfolio position.
• BTC: follow-through after the exit, especially around the 60,000 area mentioned in the order-flow post.
• MU: posted S/R-based long setup if downside rejection appears.
• GOOG: rebound was described as distribution; only puts / bearish exposure was mentioned.
• Levels explicitly posted in the risk reports:
  • ASML 1D: support 1,810–1,848; resistance 1,832–1,900
  • SPCX 4H: resistance 160.93–166.88
  • ANET: 4H support 161.51–164.28; 1D support 159.99–164.04

Overview

• In the supplied posts for 2026-06-29 to 2026-07-06 UTC, activity was concentrated on Jun 29–30.
• The week focused on semiconductor-heavy exposure and short-term tactical trades. A Jun 29 portfolio risk report showed concentrated positions in SMH (32.38%), AMAT (14.22%), and ASML (13.57%), with AI/semi exposure at 68.71% and cash/cash-like at 34.96%.
• Trading posts showed realized gains in AMZNUSDT, AMATUSDT, and MUUSDT, while ETH was discussed as a breakout attempt within a broader range context.

Key Themes

• Concentration risk management: Repeated emphasis on using support/resistance discipline rather than chasing strength, especially given heavy AI/semi exposure.
• Semiconductor bias remained strong: AMAT was highlighted positively; some storage exposure was partially taken for profit; MU was bought with 5x leverage and later closed profitably.
• New name added: COHR was bought, with a follow-up note that its 1W chart had reached a support zone.
• ETH setup: One ETH chart post called the structure “textbook” but still range-bound; later posts treated ETH as a breakout with a near-term target around 1700, and an ideal-case chart toward ~1850.
• Enterprise software watch: A market note flagged ServiceNow + Accenture’s AI risk/cybersecurity partnership as worth tracking, but not as an immediate guidance-changing event.

Notable Posts

• 2026-06-29 08:32 UTC — Public portfolio risk report:

Based on IBKR LastBusinessDay data (not intraday real-time holdings). Key figures: SMH 32.38%, AMAT 14.22%, ASML 13.57%, AI/semi 68.71%, cash/cash-like 34.96%. SMH weight was down from 36.02% to 32.38%.

• ETF constituent changes:

QQQ: MU 5.30% → 5.75%. SMH: NVDA 17.71% → 18.16%; TSM 8.73% → 9.04%.

• Trade recap screenshots:

AMZNUSDT perpetual: +28.65% AMATUSDT perpetual: +16.51% The accompanying text said AMZN stock was still being held long term.

• COHR add:

A post said “买了一些 $COHR”, followed by a note that COHR 1W touched support zone with close=356.11.

• MU trade:

MUUSDT 5x long was documented at +15.13%, then +21.50%, and later closed at +37.38%. A follow-up post marked $MU as having “超额完成”.

• ETH management:

ETH/USDC was shown near 1,619, with the comment to keep holding and look for ~1700 before closing the trade. A separate 4H chart described an ideal case near ~1850 via a limit sell.

• NOW/ACN note:

A market post highlighted ServiceNow + Accenture launching joint AI-powered services for risk/cybersecurity workflow migration, and placed NOW / ACN on the watchlist.

Watchlist

• SMH / AMAT / ASML: Main portfolio volatility drivers; still framed around support/resistance management.
• MU: In focus after both ETF weight changes and a successful leveraged trade.
• COHR: Newly added; weekly support zone was specifically flagged.
• ETH: Watching whether the breakout extends or price remains in a broader range.
• NOW / ACN: With related comparison names also mentioned: IBM, CRM, MSFT, ORCL, PANW, CRWD.

Overview

• Supplied posts were concentrated on Jun 22–25 and split across three buckets: public portfolio risk logs, AI infrastructure/news radar, and discretionary chart/trade notes.
• The public risk reports explicitly used IBKR Flex `LastBusinessDay` data, so they were not intraday-real-time.
• Across the three portfolio updates, the reported mix moved sharply toward AI/semi exposure:
  • Cash / cash-like: 62.41% → 51.09% → 31.10%
  • AI / semiconductor exposure: 36.05% → 47.84% → 66.60%
• Reported top exposures shifted from SMH / MSFT / NVDA on Jun 22 to SMH / MSFT / MU on Jun 23 and SMH / ASML / MSFT on Jun 24.
• A separate post said direct NVDA exposure was no longer being held, with exposure moved through SMH to reduce single-name risk.
• Tactical trading posts remained active: OKB was sold at 80 against resistance, while BTC posts focused first on 65,500 resistance and later on a range/VAH reclaim as the condition for a lower-timeframe trend reversal.

Key Themes

• Semiconductor concentration increased.

SMH stayed the core vehicle, while the latest risk report also highlighted ASML, MU, AMAT, MRVL and other semi-linked names. SMH constituent updates showed a higher NVDA weight, and QQQ updates added/raised several AI-linked names.

• AI infrastructure research centered on power, land, water, and capacity.

Major posts tracked Oracle’s quantified infrastructure commitments, Microsoft/Chevron Project Kilby, TPL’s land/water role, PRIM energy awards, HIVE site buildout, and Fervo/PNNL/NVIDIA geothermal validation.

• Memory/HBM remained a major branch.

Posts upgraded the Micron/Anthropic relationship, flagged Micron Q3 AI-memory/data-center revenue, and added SK hynix’s F-1 / planned SKHY ADS listing to the watchlist.

• Execution stayed chart-driven and repeatable.

Attached screenshots showed realized/performance snapshots in AMAT, ASML, KLAC, MU, CRWV perpetuals, with an accompanying note that the setup logic was broadly the same across names.

• Secondary enterprise-budget read-throughs were also tracked.

Posts covered IBM/OpenAI in application security, Coursera/Udemy combined modeling, Korn Ferry talent demand, and Paychex FY26/Q4 plus its WISE AI engine.

Notable Posts

• Jun 22 – Fervo / PNNL / NVIDIA:

EGS-Twin geothermal digital twin work was framed as a technical-validation signal, not a new backlog/order.

• Jun 22 – Oracle FY2026 10-K:

The post highlighted $13.309B of data-center power-related obligations and $19B of additional cloud infrastructure purchase commitments.

• Jun 23 – Microsoft / Chevron Project Kilby:

A 20-year PPA targeting about 2.67GW for West Texas data centers was cited, with GE Vernova and Caterpillar Solar Turbines named in the read-through.

• Jun 23 after close – TPL and Cerebras:

TPL was linked to Project Kilby through land + brackish water resources; Cerebras posted its first public-company Q1 update and reiterated the OpenAI >$20B / 750MW and AWS branches.

• Jun 24 – GPUS Michigan MSA:

GPUS announced, and later filed via 8-K / MSA exhibit, an initial 20MW AI-compute deal with potential >$1.2B maximum-term revenue and expansion to 52MW / >$3.0B. Posts repeatedly flagged customer anonymity and execution risk.

• Jun 24–25 – Micron / SK hynix / semi trades:

The feed upgraded the Micron–Anthropic supply relationship, logged Micron Q3 AI-memory/data-center strength, added SK hynix F-1 / SKHY to the IPO watchlist, and shared Binance screenshots showing AMAT +37.62%, ASML +15.37%, KLAC +13.36%, MU +89.96%, CRWV -4.26%.

• Jun 22–25 – Crypto charts:

A post said OKB bought below 73 was sold at 80 due to overhead pressure; BTC posts first marked 65,500 as resistance and later said the market needed to build above VAH to confirm a lower-timeframe reversal.

Watchlist

• Portfolio risk drift: whether the lower reported cash buffer and higher AI/semi concentration persist in the next lagged Flex reports.
• Project Kilby chain: Chevron FID, GEV/CAT order disclosure, TPL land/water economics, and permitting/interconnection progress.
• GPUS: delivery against the 20MW MSA, deposits/upfront payments, customer identity, retrofit financing, and supplier disclosures.
• Memory / HBM: more detail on Micron’s Anthropic agreement, Micron strategic customer agreements, and SKHY filing/pricing/listing progress.
• Power/infrastructure conversion: whether Oracle, Fervo, PRIM, HIVE updates translate into disclosed project-level PPAs, MW, supplier awards, or revenue.
• Crypto levels: whether BTC can hold/reclaim the posted range VAH, and whether OKB revisits the 80 resistance area.

Overview

• In the supplied posts for this window, activity was concentrated on 2026-06-15 to 2026-06-16.
• The overall tone was risk-aware rather than outright bearish: a broad 4H scan flagged many names at resistance, with the explicit reminder: “不是看空，只是提醒及时落袋为安而已”.
• Portfolio updates showed a move toward more cash and lower AI/semi concentration, while crypto commentary focused on BTC breakout/value-area behavior and ETH support tests.
• Research posts stayed active across AI infrastructure, AI workflow, and event-driven listings/orders.

Key Themes

• Take-profit / no-chase discipline:

A 4H radar scan flagged RKLB, IONQ, NVDA, CRWV, MSFT, ASTS, DUOL, CRWD, PLTR, ORCL, MRVL, APLD, META, FUTU in resistance zones; SNPS was flagged at 4H support. Several posts repeated a “watch, don’t chase” stance.

• Portfolio de-risking:

On 6/15, disclosed concentration was led by NVDA 17.64%, AVGO 14.60%, SMH 11.84%, MSFT 11.20%, with AI/semi exposure 54.22% and cash 50.10%. By 6/16, AVGO 19.30% and MSFT 14.69% were the largest positions, while NVDA fell to 13.01%, SMH to 7.93%, AMAT to 0.06%, AI/semi exposure fell to 46.51%, and cash rose to 60.78%.

• Crypto market-structure focus:

BTC posts centered on breakout confirmation, sell-order absorption, re-accumulation near resistance, and VAH/POC acceptance. ETH was first noted as a missed stronger squeeze, then later as retesting support, with 1720 mentioned as the next level if support fails.

• Research-driven AI infra tracking:

The week included multiple event notes on LIXT/NOMAD, SPCX, AIBZ, GPUS, TSEM/IQE, SHMD, SMCI, and CRM/Fin.

• Narrowing focus:

There was also an explicit decision to track fewer names, especially in US equities, and focus more on mainstream tickers.

Notable Posts

• 6/15 Portfolio risk report:

Heavy AI/semi exposure remained, but the framework emphasized support/resistance discipline and avoiding aggressive chasing.

• 6/15 MSFT / BTC notes:

MSFT was described as a prior-week buy around 386. BTC spot was partially reduced into a watched breakout area. A later BTCUSDT perpetual screenshot showed a long at +17.86%, with open 65,931.2 and latest 67,139.4.

• 6/15 AI infra/event updates:
  • LIXT / NOMAD: acquisition framed as a high-risk early signal tied to mobile BESS and AI/data-center power bottlenecks, not a confirmed AI order.
  • SPCX: actual Nasdaq trading was confirmed.
  • AIBZ / GPUS / TSEM-IQE: new capacity/supply-chain signals were logged, with different confidence levels.
  • SMCI: the $39B AI server orders did not increase, but a 75M depositary share offering closed for about $3.75B gross proceeds.
  • CRM: Salesforce signed a deal to acquire Fin for about $3.6B.
• 6/16 Portfolio shift:

Cash increased, AI/semi exposure decreased, and AVGO became the largest disclosed position.

• 6/16 SHMD / market notes:

SHMD disclosed more than €26M of order intake since mid-May tied to AI infrastructure and optical-module demand. Separate tactical posts noted a missed AMKR move, no desire to force trades, BTC waiting for post-break acceptance, ETH on support, and continued patience on SMH.

Watchlist

• Core portfolio names: AVGO, MSFT, NVDA, SMH, ORCL, ANET, AMZN, ISRG — mainly for position sizing and resistance/support discipline.
• Names flagged near resistance/support: RKLB, IONQ, CRWV, ASTS, DUOL, CRWD, PLTR, MRVL, APLD, META, FUTU; SNPS for support-zone behavior.
• Crypto:
  • BTC: whether price can hold above the new value area / VAH after breakout.
  • ETH: whether support holds or opens the path toward 1720.
• AI infra / event names: LIXT/NOMAD, SPCX, AIBZ, GPUS, TSEM/IQE, SHMD, SMCI, CRM.

Also noted: QMLS still showed no confirmed trading, and ZFIV had not been observed as actually trading.