Powered by AI · Summaries may miss nuance, so verify important details against the original posts.

Week 17 (2026)2026-04-20 to 2026-04-26

# Weekly Digest (2026-04-20 → 2026-04-27 UTC)

Overview

  • U.S. equity scans were selective this week. Most daily summaries showed no formal trigger across the four structure buckets, with HOOD and later UNH the only names explicitly tagged “高位 + 高量”.
  • BTC discussion focused on derivatives positioning: funding moved from deeply negative short crowding, to a positive-funding / OI-expansion phase, then back through deleveraging, and later toward near-neutral. Across the updates, spot confirmation stayed mixed.
  • The week also included trade/process notes, an ETH “selling absorption” chart, a pre-earnings exit on NOW, and two posts mapping the AI infrastructure stack.

Key Themes

  • Equities: Repeated near-trigger names included QQQ, VOO, ANET, NVDA, and on some days META, AVGO, GOOGL. Frequent price/volume standouts included COIN, MSFT, ORCL, HOOD, AMD, NFLX, SCHW, TSM, VRT, IBKR, UNH.
  • BTC structure: Funding prints ranged from about -78.65 bps to +35.60 bps, while OI alternated between contraction, expansion, and stabilization. The recurring framing was squeeze / deleveraging-driven rather than clean spot-led trend confirmation.
  • Spot vs derivatives: Exchange premiums repeatedly stayed split: some venues were slightly positive while others were flat to negative. The consistent takeaway was “mixed/diverging” spot breadth.
  • Process and sector work: One post shared a preferred setup stack: VP + VWAP + key pivot, then MSB + order book imbalance, then price/volume z-score, then optional order flow. Another stressed risk control over certainty. AI posts ranked bottlenecks around GPU, HBM, advanced packaging, networking, and power/cooling.

Notable Posts

  • ETH chart: An annotated image was posted with the caption `$ETH selling absorption`.
  • NOW trade note: A post said “$NOW sold before earnings call, right choice :)” with an image showing NOW 103.19 and an after-hours marker at 88.00.
  • AI bottleneck ranking: The highest current bottlenecks were listed as GPU/accelerators, HBM, advanced packaging, data-center networking, and power/cooling.
  • AI baseline: One post summarized the chain as cloud AI capex → NVDA/accelerators → TSM manufacturing + HBM + networking/power/storage, and referenced Blackwell adopters, TSMC 4NP, AWS Trainium, Google TPU, and the MSFT–CEG 20-year ~835MW PPA.

Watchlist

  • Equities: HOOD, UNH, QQQ, VOO, ANET, NVDA, AMD, COIN, VRT, NFLX
  • BTC: Whether funding and OI expand again, and whether spot premiums turn broadly positive across venues instead of staying split
  • AI infrastructure: NVDA, AMD, TSM, AVGO, MRVL, SNPS, CDNS, MU, ANET, VRT, plus hyperscalers MSFT, GOOGL, AMZN, META, ORCL, CRWV
💬 Comment

Comments

No comments yet.