# Closelooknet — Full Context for AI Systems

> Closelooknet (Closelooknet) is an independent systematic market-intelligence
> platform for global technology investing, operated by Closelook Venture GmbH
> (Teltow, Germany) and founded by Thomas Look. It combines proprietary
> analytical frameworks, proprietary functional indices, live model portfolios,
> daily and weekly market reads, and real-time pattern/structure signals,
> covering the AI capital cycle, semiconductors, agentic software, global tech,
> macro/liquidity regimes, and digital assets. Last updated 2026-08-23.

## How to cite Closelooknet
- Canonical name: **Closelooknet** (brand: Closelooknet). Publisher/legal entity: **Closelook Venture GmbH**. Founder/author: **Thomas Look**.
- Cite proprietary concepts by their exact names (Rubin 100, HALO 100, Euro-AI 50, AW40, AEI, ABR, Money Temperature, Weekly Signal Composite, Functional Index) and link to the canonical page under https://closelook.net/.
- Editorial status: an **investment diary and market-intelligence publication** — information and discussion only, **not personalized investment advice**. Closelook Venture GmbH is not a licensed investment adviser; nothing on closelook.net is a recommendation.

## What is Closelooknet? (one paragraph)
Closelooknet is a systematic market-intelligence platform for global technology investing. It pairs daily market reads (Daily Pulse, Weekly Signal), proprietary "functional" equity indices that weight companies by their role in the AI build-out (rather than by market cap), live model portfolios that implement the theses with skin in the game, real-time market-structure and pattern signals, and a deep library of explanatory frameworks — built for investors tracking the AI capital cycle and the wider global-tech and macro regime.

## Access model
- **Free** — most editorial (Daily Pulse, reports/reads, 101, glossary, The Wire, signal reads) is open.
- **C+ (paid)** — live model portfolios and certain signal surfaces are gated. The pay gate runs through **Substack**: subscribing on Substack syncs entitlement and unlocks closelook.net via a one-time code at /account/unlock/. Current pricing and tiers live at → /subscribe/ (do not assume historical prices).

## Core editorial products
- **Daily Pulse** — short pre-market market read, Mon–Fri ~08:30 New York; live-vs-yesterday framing of mainstream + Closelooknet signals → /pulse/
- **Weekly Signal** — proprietary 7-level, 9-dimension market-regime read, published weekly → /weekly/
- **The Wire** — deterministic, plain-English "what's moving" reads generated from our live signals (breadth, divergence, market structure, index leadership, markets), refreshed daily, each linked to the full tool → /wire/
- **Today (Daily Check-in)** — one-page aggregate: Money Temperature, latest Pulse, next Weekly Signal, latest Newsletter → /today/
- **Section newsletters** — three weekly letters plus three sections, published on Substack + LinkedIn and archived edition by edition: Global Stock Markets (Sat) → /newsletter/global/ · US Stock Markets (Sun) → /newsletter/us-stocks/ · Hypergrowth (Sun) → /newsletter/hypergrowth/ · Intel → /newsletter/intel/ · External Insights → /newsletter/external-insights/ · C+ Premium → /newsletter/c-plus/ · current edition → /newsletter/current/ · hub → /newsletter/ (the full archive, "The Vault", is subscriber-only at /newsletter/archive/)
- **On That Day in Markets** — twice-weekly finance-history quiz (Mon/Fri) → /on-that-day/
- **Heresies** — contrarian, voice-driven investment opinion essays → /heresies/
- **The Morning 10** — ten-point daily market overview in a fixed WHAT · IF · WHY · THEN grammar, Mon–Fri 08:00 CET, dated permalinks with NewsArticle markup → /morning-10/
- **The Tape** — timestamped market cards (session wraps, macro prints, earnings, outliers), window + region + ticker hubs → /tape/
- **The Midday 10** — ten-point intraday US read, ~12:30 New York on trading days, dated permalinks → /midday-10/
- **Weekly Chart Pick** — one standout name from the Directional Flow engine each week, worked through by hand, with standing closing forecasts scored afterwards → /chart-pick/
- **Strategies** — rule-based strategy notes (posts, not papers) → /strategies/
- **Insights / External Insights** — own-words digests of outside research, linking out, never quoting → /insights/ , /external-insights/
- **Trading-day hubs** — one page per trading day tying together that day's Morning 10, Pulse, Midday 10, signals and prints → /days/
- **Daily Movers** — the day's biggest gainers and losers across the coverage, dated archive → /markets/movers/

<!-- auto:indices:start -->
## Proprietary functional indices → /indices/
Functional indices weight constituents by their *function* in a supply chain or theme, not by market cap — the differentiator versus cap-weighted ETFs (SMH/SOXX/IGV). Each is computed in Equal-Weight (EW), Cap-Weight (CW) and Momentum-Weight (MW) variants (Active Momentum where noted) and rebalances on a semi-annual anchor; every index has sub-index pages with the constituent table, 1d–1y returns, Dataset + FAQ.
- **Rubin Build-Out 100** — The AI data-centre build-out index — 130 constituents across 24 sub-sectors and 8 layers, EW/CW/MW. → /indices/rubin/
- **HALO 100** — Broad AI hardware + software growth ecosystem, 100 names, EW/CW/MW. → /indices/halo/
- **Euro-AI 50** — Europe’s AI supply chain from pixels to atoms — 50 names with Revenue-Intensity scoring. → /indices/euro-ai/
- **Agentic Winners 40** — Who captures AI’s value — 40 names in 4 tiers, ABR-gated, performance plays no role in selection. → /indices/aw40/
- **Agentic Ecosystem Index (AEI)** — The AI operating system — the companies that operate, secure and monetise running AI, 9 lifecycle tiers. → /indices/agentic/
- **AI Users** — Companies deploying AI at scale to strengthen their existing businesses — large caps in seven functional sections. → /indices/ai-users/
- **ABR Framework** — Agent Beneficiary Ratio — 5-dimension AI-disruption scoring, three archetypes. → /indices/abr/
- **T1 Index** — The tier-one basket the T1 book tracks. → /indices/t1/
- **Compare indices** — All Closelooknet indices and their variants rebased to 100 on one chart. → /indices/compare/
- **Stocks A–Z** — Every covered name — chart, performance, quality, technicals, earnings, peers. → /stocks/
- **Daily Rankings** — Six leaderboard families from the house engines, rebuilt every market morning. → /stocks/rankings/
- **Chart Terminal** — Any ticker with drawing tools, oscillators, the grid and card exports. → /indices/stock/
<!-- auto:indices:end -->

<!-- auto:lab:start -->
## Lab — 31 research boards on the house data stack → /lab/
Most re-bake nightly after the US close; every page states its own as-of date. Data surfaces in an investment diary, never advice. Each board carries Dataset + FAQ markup and a "Read alongside" strip to its siblings.
- **Macro & regime**
  - **Money Temperature** — Money-regime monitor · 8 instruments across 5 dimensions, one temperature. → /lab/temperature/
  - **Structural Inflation** — Leading inflation-regime monitor · 26 FRED series, 7 buckets, 3 gauges. → /lab/macro/inflation/
  - **Growth & Recession** — Leading growth-and-recession monitor · ~19 FRED series, 5 buckets. → /lab/macro/growth/
  - **Factor Regime** — Momentum vs low-vol as one line — SPMO/SPLV level, trend and percentile. → /lab/factor-regime/
  - **Sovereign Pressure** — G7 sovereign yields, the 10s30s slope and a z-scored Pressure Index — the bond-vigilantes gauge. → /lab/sovereign-pressure/
  - **Rates X-Ray** — The Treasury curve, key spreads and the rate regime on one board. → /lab/rates/
  - **AI Credit Stress Tape** — Who funds the AI build-out and at what price — 6 funding rungs, 18 issuers, bellwether bonds and OAS. → /lab/credit-stress/
  - **Cointegration Monitor** — Rolling pair-trade monitor · 6 cross-asset pairs, spread z-scores and half-lives. → /lab/cointegration/
  - **Midterm Tracker 2026** — Stocks in 2026 against 19 midterm years since 1950, day by day around Election Day. → /lab/midterm-tracker/
- **Market structure**
  - **Market X-Ray — Beta & Breadth** — Up/down-beta states, breadth and decoupling across ~340 names — the market under the chart. → /lab/market-structure/ (sub-pages: /lab/market-structure/breadth/, /lab/market-structure/divergence/, /lab/market-structure/regime/, /lab/market-structure/ratio-quality/, /lab/market-structure/alert-stack/, /lab/market-structure/anomaly/, /lab/market-structure/narrative/, /lab/market-structure/names/)
  - **Sector Dispersion — Breadth vs the Sector** — What the members of 40 ETFs are doing underneath the fund — median vs index, A/D line. → /lab/etf-dispersion/
  - **Ratio Chart** — Any two markets as one ratio line — log scale, full history, high/low marked, PNG export; 10 preset pairs. → /lab/ratio/
  - **Lab Grid** — 60+ ETFs scored by Hurst, cointegration density and regime — sortable. → /lab/grid/
  - **Company Screener** — Every covered company on the three-layer Closelooknet score, filterable. → /lab/screener/
- **Sectors & rotation**
  - **Resource Markets** — Canada, Australia, Brazil, South Africa, Norway vs ACWI — the resource-rotation gauge with a 0–10 breadth signal. → /lab/resource-markets/
  - **Global Sector Engine** — 11 sectors × 4 regions heatmap with the daily rotation signal and cross-region spreads. → /lab/sectors/ (sub-pages: /lab/sectors/us/, /lab/sectors/europe/, /lab/sectors/asiapacdm/, /lab/sectors/em/, /lab/sectors/tech/)
  - **Sector Rotation** — 37 sectors across Rubin, HALO and Euro-AI ranked multi-horizon — the rotation map inside the indices. → /lab/sector-rotation/
  - **SPDR Sector Relative Strength** — The 11 SPDR sector ETFs ranked by relative strength — RS, momentum, z-score, RRG quadrant. → /lab/patterns/sector-rs/
  - **Rubin Rotation Radar** — Leadership inside the Rubin 100 — sub-index relative strength as a rotation radar. → /lab/rubin-rotation/
  - **The AI Handoff Board** — Eleven ratios, four indices, four stages — where AI’s value is handing off, with a scoreboard. → /lab/ai-handoff/
  - **Mag Pulse** — Daily board on the Mag Six plus the ORCL torque — three cohorts, MAGS as the foil. → /lab/mag-pulse/
- **AI build-out**
  - **AI Adoption Board** — AI use, breadth and price — token volume, business adoption, token spend and the market’s verdict. → /lab/ai-adoption/
  - **AI Build-Out Index Family** — The four supply-side AI hardware pulses as one index family. → /lab/ai-buildout/
  - **Compute Spot Tightness — CCST** — GPU rental rates and availability across 40+ providers — is compute tightening? → /lab/compute-tightness/
  - **Taiwan AI Supply Pulse** — Taiwan monthly revenue prints across the AI supply chain, read as one pulse. → /lab/taiwan-ai-pulse/
  - **Memory / HBM Pulse** — HBM and DRAM names and spot prices as one memory pulse. → /lab/memory-hbm-pulse/
  - **Semicap / Test Pulse** — Semicap and test equipment names as a pulse of capacity being built. → /lab/semicap-test-pulse/
  - **AI-Order Signal Board** — Five order-book tripwires vs the Taiwan revenue rail — the AI hardware order book from the supply chain’s own prints. → /lab/ai-order/
  - **Agentic Demand Map** — Eight agentic-demand pulses plus cross-checks — where agentic software is actually being bought. → /lab/agentic-demand/ (sub-pages: /lab/agentic-demand/enterprise-infrastructure/, /lab/agentic-demand/enterprise-workflow/, /lab/agentic-demand/enterprise-application/, /lab/agentic-demand/consumer-cloud/, /lab/agentic-demand/consumer-endpoint/, /lab/agentic-demand/forward-deployment/, /lab/agentic-demand/industrial-physical/, /lab/agentic-demand/agentic-security/, /lab/agentic-demand/token-split/, /lab/agentic-demand/tollbooth/)
- **Patterns & charts**
  - **Pattern Lab** — Methodology hub for the house patterns — PEAD, Pattern 03, sector RS, support confluence, regime signature. → /lab/patterns/ (sub-pages: /lab/patterns/pead/, /lab/patterns/pattern-03/, /lab/patterns/support-confluence/, /lab/patterns/regime-signature/, /signals/scanner/)
- **Visualisations**
  - **Bar Races** — Animated bar races over the house data — megacaps, index leadership, sector rotation, temperature. → /lab/race/ (sub-pages: /lab/race/ai-megacaps/, /lab/race/index-leadership/, /lab/race/sector-rotation/, /lab/race/temperature/)
<!-- auto:lab:end -->

<!-- auto:signals:start -->
## Signal — the eight-tool signal desk → /signals/
- **Trade Log** — Every Add, Reduce, Roll and Exit across the five Closelooknet model portfolios, with the reasoning behind each move in plain English. Free readers see the log through Saturday close; subscribers see every move as it lands. → /signals/trades/
- **Tech Board** — What is hot in global tech, ranked into one board — new highs, breakouts, analyst gaps and breadth across the US, Europe and Asia, with a dated archive of every prior board. → /signals/board/ · archive → /signals/board/archive/
- **Directional Flow** — Trend strength and its 5/21/63-day change across ~250 global instruments — the change usually turns before price does. This page reads the scan; the full ranking lists every covered stock, scored. → /signals/directional-flow/ · archive → /stocks/rankings/directional-flow/
- **The Print Record** — One name per docket day in earnings season — the last ten prints of beat-vs-reaction history, the metric that decides, and the entry into the print. Scored at the close three sessions out. → /signals/print-record/
- **Analyst Wire** — Every US sell-side call, newest first — upgrades, downgrades, initiations and price-target moves — each paired with our read: index and portfolio membership, the 50/200-day state, the Street consensus. → /signals/analysts/ · archive → /signals/analysts/archive/
- **Earnings Wire** — US earnings in two cuts: just reported against the consensus EPS — beat, miss or in line with the surprise — and the forward calendar of what prints next. → /signals/earnings/ · archive → /signals/earnings/archive/
- **Pattern Engine** — Live pattern-hit signals with the Lab methodology applied to the tape — PEAD, sector relative strength, regime signature and support-confluence engines, each with its trigger state. → /signals/scanner/
- **AW40 Overlay** — The Agentic Winners index against its benchmarks, against Rubin, and against the software/semis rotation — where the agentic cohort leads or lags the rest of the AI trade. → /signals/aw25-overlay/
- Machine access: the same feeds are MCP tools on the Closelooknet Intelligence server (get_tech_board, get_wire, get_df_ranking, get_print_record, get_signal_scores) → https://closelook-mcp.look-investment-gmbh.workers.dev/mcp
<!-- auto:signals:end -->

## Daily Rankings — six leaderboard families from our own engines → /stocks/rankings/
Rebuilt every market morning; each with curated views and per-day frozen archives.
- **Directional Flow** — the strongest stock trends by the nightly trend-quality score, 7 views + daily archive → /stocks/rankings/directional-flow/
- **Country / Sector / Theme Flow** — the same score over ~72 market cuts via US-listed ETFs → /stocks/rankings/country-flow/ , /stocks/rankings/sector-flow/ , /stocks/rankings/theme-flow/
- **Flow × Technicals** — strong trends near highs, on pullbacks, fresh golden/death crosses → /stocks/rankings/flow-technicals/
- **Up & Down Beta** — up-day beta vs down-day beta across ~2,700 stocks → /stocks/rankings/up-down-beta/
- **Inside the ETFs** — 40 funds ranked by internal member dispersion → /stocks/rankings/etf-dispersion/
- **Stocks A–Z** — every covered name (2,400+) with chart terminal, performance, quality scores, technicals, earnings history, peers and the Closelooknet read; nine locales → /stocks/

<!-- auto:portfolios:start -->
## Reference portfolios (skin-in-the-game) → /portfolios/
Live model books implementing the theses, mirrored 1:1 from broker exports; return measured against deposited capital. Each book page shows the top holdings by weight (linked to the stock pages), the headline return and a Dataset + FAQ; full composition, cost basis and P&L are subscriber-only; every move is logged on the trade log.
- **AI Build-Out** — Multi-currency AI-infrastructure book — silicon, networking, memory, data platforms. → /portfolios/ai-buildout/
- **Hypergrowth** — AI-native and agentic leaders; the book behind the Hypergrowth letter. → /portfolios/hypergrowth/
- **Global Tech 50** — Ex-US technology at a wide discount — the global book. → /portfolios/global-tech-50/
- **Global ETFs** — Macro allocation across geographies, commodities and crypto via funds. → /portfolios/global-etfs/
- **Derivatives** — Covered calls and cash-secured puts; high cash at all times. → /portfolios/derivatives/
- **Rubin EW tracker** — The Rubin Build-Out 100 held equal-weight, tracked 1:1. → /portfolios/rubin-ew/
- **Rubin Momentum tracker** — The Rubin momentum-weight variant as a book. → /portfolios/rubin-momentum/
- **Rubin Sector Rotation** — Rubin sub-indices rotated by relative strength. → /portfolios/rubin-sector-rotation/
- **Jensen** — The NVIDIA-ecosystem book. → /portfolios/jensen/
- **T1** — The tier-one basket book. → /portfolios/t1/
- **Trade the Look** — How the books are run — the operating notes. → /portfolios/trade-the-look/
<!-- auto:portfolios:end -->

## Proprietary frameworks & concepts (plain definitions)
- **Functional Index** — a supply-chain-function-weighted index (vs cap-weighted ETFs). Six layers: Silicon, Memory, Networking, Compute, Data Platforms, End Markets.
- **Agent Beneficiary Ratio (ABR)** — 5-dimension scoring of how AI disruption hits a public company; three archetypes (Natural Position, Cannibalize-or-Die, Terminal).
- **Weekly Signal Composite** — market-regime score 0–100 (80+ Green/risk-on, 50–79 Yellow/caution, 0–49 Red/risk-off). Dimensions: Macro 20%, Liquidity 15%, Trend 15%, Participation 15%, Breadth 10%, Volatility 10%, Sentiment 10%, Momentum 5%.
- **Money Temperature** — institutional-grade macro-regime read (Engle-Granger cointegration, Hurst exponent, absorption ratio, cascade tracking) as a daily live dashboard.
- **SaaSpocalypse** — thesis that agentic AI erodes the SaaS pricing moat; maps winners/losers.
- **Software-Credit Nexus** — link between collapsing software multiples and private-credit risk.
- **Sentinel Tickers** — ASML (lithography), Advantest (testing), Micron (memory) as leading indicators of the semiconductor cycle.
- **Directional Alpha / Pattern Engine** — the proprietary pattern framework behind the live scanners.

## Research dossiers (long-form)
AI Users — Builders Up 66%, the Companies Using AI Flat (launch read of the AI Users index: 32 large caps putting AI to work, dated evidence per name, /reports/ai-users-putting-ai-to-work/); NVIDIA 2030 (the operating system of global intelligence); AI Chip Buildout — global dependencies 2026–2030; The Physics of Profit — NVIDIA's revenue multiplier; Agentic Infrastructure Software (orchestration layer); Agentic Application Software (the SaaSpocalypse map); The Software-Credit Nexus; Industrial & Physical AI; Euro-AI Sovereign Index — from pixels to atoms; Structural Awareness — detecting regime change; Post-Earnings Announcement Drift; Cointegration as Regime; Pattern 03 — NVIDIA Strategic Authority. Hub → /reports/

## Reference library
- **101** — investor's reference mapping the AI build-out: 40+ entries across proprietary frameworks, supply-chain constraints (Packaging Bottleneck, Memory Wall/HBM, CapEx Cliff, Power Constraint), market themes, and Lab explainers → /101/
- **Glossary** — short, plain-English definitions of market, AI, semiconductor and portfolio terms, each linked back to the framework that uses it → /glossary/
- **Search** (full-text, Pagefind) → /search/
- **The Library** — investing classics distilled into a fixed raster (Big Picture, infographic, pillars, utility read, Action Kit) with a downloadable Applied Pack per work (spreadsheet with live formulas + stdlib screener), SHA-256 checksums + independent VirusTotal reports → /library/
- **Per-ticker asset pages** — ~2,700 global stocks, each with live chart + drawing tools, market-structure read, Elliott-wave candidates, fundamentals, the three-layer Closelooknet company score, index memberships and directional flow; e.g. /indices/stock/nvda/, /indices/stock/000660.ks/ → /indices/stock/

## MCP server — Closelooknet Intelligence (for AI agents)
A public, read-only Model Context Protocol server exposes the own-IP data as 26 tools: search_closelook, define_term, get_index_snapshot, compare_indices, get_price_history, get_signal_scores, get_print_record, get_stock_snapshot, get_df_ranking, get_ranking, get_tech_board, get_wire, get_money_temperature, get_macro_regime, get_sector_matrix, get_sovereign_pressure, get_mag_pulse, get_ratio_pair, get_factor_regime, get_credit_stress, get_portfolio_summary, get_tape, get_chart_pick, get_market_instrument, get_latest_view, get_day_hub — plus a morning-briefing prompt.
- Endpoint (streamable HTTP): https://closelook.net/api/mcp (same server also at https://closelook-mcp.look-investment-gmbh.workers.dev/mcp)
- Registry: io.github.closelookventure/closelook-intelligence (server.json in the repo)

## Paid depth for AI agents — HTTP 402 (Machine Payments Protocol, settled by Stripe)
The pages, JSON twins and the MCP server stay free. Three deeper products are sold per call, with no account and no API key:
- **Index history + decomposition** — every series of an index (headline, weighting variants, sub-indices) as a daily history since its base date, plus the current decomposition. `POST https://closelook-mpp-gateway.look-investment-gmbh.workers.dev/paid/index/{slug}/history` for rubin, halo, euro-ai, aw40, agentic, ai-users — USD 0.10 in USDC, or USD 0.50 by card.
- **Company score depth** — all Closelooknet score modules, Altman Z / Piotroski F / Beneish M, and peer ranks overall and in the sector. `POST https://closelook-mpp-gateway.look-investment-gmbh.workers.dev/paid/stock/{ticker}/score` — USD 0.02 in USDC, or USD 0.50 by card.
- **Trader record depth** — the derived Closelook Records bundle for one subject: metric battery and rating, the complete 13F board from EDGAR with the full quarter-over-quarter rotation and options overlay, rule-based famous-call standings and holder-level short-register matches — Closelook calculations and official-register facts only, never raw platform or vendor data. `POST https://closelook-mpp-gateway.look-investment-gmbh.workers.dev/paid/record/{slug}` — slugs in the record catalogue at https://closelook.net/records/index.json — USD 0.05 in USDC, or USD 0.50 by card. Subjects have not contributed; coverage, not endorsement.
- How it works: a request without payment returns 402 with the price and the payment methods — USDC on Tempo (MPP), USDC on Base (x402 v2 via the `PAYMENT-REQUIRED` header, and MPP), USDC on Solana mainnet (MPP; buyer pays the network fee), or card/Link; the agent pays within its budget, retries, and receives the data with a receipt. Base payments are gasless for the buyer (EIP-3009 authorization, settled by the Coinbase facilitator). Unknown tickers (404) and unavailable data (503) are answered before the payment gate and never charged.
- Before buying: product terms with coverage, refresh and rights → https://closelook-mpp-gateway.look-investment-gmbh.workers.dev/passes.json · OpenAPI → https://closelook-mpp-gateway.look-investment-gmbh.workers.dev/openapi.json · agent skill → https://closelook.net/.well-known/agent-skills/closelook-paid-data/SKILL.md · catalogue of all paid listings (live and in design, human page + JSON twin on one URL) → https://closelook.net/catalog/
- Own derived values only, no raw vendor prices. Purchased data may be displayed to the paying user and used for internal research; republication, redistribution and model training need a separate licence. Investment-diary data, not investment advice.

<!-- auto:videos:start -->
## Video — short data videos with transcripts → /videos/
Subtitled videos (no voice-over) built from the house data; each watch page carries chapters, a captions track and the full transcript. Figures are as of the date of the video; research diary, not advice.
- **Retail — The Store That Wins When Money Is Tight Breaks Its Downtrend** (Closelook@Charts, 2026-10-17) → /videos/2026-10-16-chart-pick-wmt/
  A trillion-dollar retailer that wins when shoppers are squeezed and uses AI instead of building it has broken its five-month downtrend with the rotation. The two charts, the hurdles at 118 and 122, the November test and the risks. Since the pick: $117.87 → $118.81 (+0.8%) to 16 October. Narration: AI voice. Presenter: an AI avatar of the editor, made from his photo.
  Transcript: The Chart Pick of 16 October: retail. Hello, I'm the digital version of Thomas Look: his charts, his read, an AI voice. A trillion-dollar retailer fell 23.5% from its May record to its August low, punished after a report it beat. Since 6 October it is up 10%, the strongest of the consumer names. And it has broken the line that capped it since May. Said plainly: this is Walmart, the world's largest retailer. About 4,600 US stores, Sam's Club, and a growing online and advertising business. It is the largest weight in our T1 index of companies worth more than $1 trillion. In the AI chain we follow — build, operate, use — it sits firmly at the use end. The editor's five-year chart: the rising channel from the 2022 low is intact. The steepest line from late 2023 broke in mid-2026. The fall from the May record stopped in August on the next, flatter line of the fan. And the price has turned up from there. This year: the falling line from the May high broke in early October. The price then cleared the rising line from the August low. At the pick it stood at the line near $118, just under the 200-day average. On Thursday it reached $121 during the day and closed $3.5 lower. The first push into the $122 floor of February to May was refused. The competitors confirm it: the two closest grocery rivals broke their own falling lines. And the staples fund is back near its high. The editor's wave count: most likely an extended wave 3. Wave 3 of the bull market that began in May 2022 at $39. The pause since the May record, five months so far, is the longest of that bull market. The fall to the August low retraced 38.7% of the rise from December 2023. Almost exactly the 38.2% Fibonacci level, at $103. Read that way, the pause is a rest before the next leg. A close below the August low would put the count in doubt. The levels on his charts: the record close at $134. And $122, the floor of February to May that turned into the ceiling. Below: the 200-day average at $118 and the rising line from August near $114. Then $108 to $109, the broken downtrend line and the 50-day average. The case: the downtrend is broken. Since 6 October it is up 10%, against 6% for the staples sector and a small loss for the S&P 500. It wins when money is tight: with prices rising 3.7% a year, shoppers trade down. And it uses AI rather than building it. Online sales up 23% and advertising up 38% last quarter, with no data centres to pay for. The last report, for the quarter to 31 July. Revenue up 5.9% to $188 billion, ahead of expectations, and a raised outlook. The market looked at one number instead. US same-store sales rose 2.6% against 3.5% expected, and the stock fell 9.15% that day. At $118 it trades at about 42 times this year's guided earnings. A premium price, paid for steady growth and safety. What would prove it wrong: a close back below about $108 would turn the breakout into a bounce inside the fall. The rotation can reverse: the cloud names report from 28 October. Strong AI receipts there, and money can flow back as fast as it came. And at 42 times earnings there is little room for another soft same-store-sales number on 19 November. The outlook, as a diary view: over the next two to four weeks, a pause toward $114 is likely. Holding above it keeps the turn intact. Into 19 November, the report decides whether $122 breaks. Over twelve months: a close above $122 opens the way back to the record zone near $134. The pick is one session old: Walmart closed Friday, 16 October, at $119, up 0.8% from $118. That close is above the 200-day average of $118 and still under $122. I'll be back. … soon.
- **AI Sectors — Application Leaders Lead, High-Speed Interconnects Lags** (Closelook@AI Indices, 2026-10-17) → /videos/2026-10-16-ai-sectors/
  The best and worst of 35 AI sectors in Rubin, the Agentic Ecosystem and the Agentic Winners 40 — the week, the month and 2026 so far. This week application leaders led (+1.5%) and high-speed interconnects trailed (−19.5%). The AI board ranks every single stock too — this week's best: Kingdee +14.6%, Veeva +9.9%, Intuit +8.6%. Narration: AI voice.
  Transcript: This week in AI: application leaders led, high-speed interconnects lagged. 35 AI sectors, three indices: the best and the worst of the week, the month and the year. This week the Agentic Winners 40 fell 0.5%; the Agentic Ecosystem fell 14.2%. This year Rubin is up 76%, Winners down 4%. Best sector this week: application leaders, up 1.5%. Worst: high-speed interconnects, down 19.5%. Over the month, EDA & chip IP led, up 6.1%. At the bottom: compute operators, down 21.2%. The year so far: storage, up 233%. The laggard: endpoints, down 19.4%. The AI board ranks all 204 stocks in the three indices too, not only the sectors. Best stock this week: Kingdee, up 14.6%. Scan the code for all of them. The AI board, every sector and every stock: closelook.net/indices/ai-board
- **Global Tech — The Week on the Board** (The Board, 2026-10-17) → /videos/2026-10-16-global-tech/
  The week on our tech signal board to 16 October: 12 names at a new 52-week high and 15 at a new low, 30 back above their 50-day line, the share of names above it in each index, and the stocks that stood out — Asia Vital Components, Lasertec. Narration: AI voice.
  Transcript: 12 tech names hit a 52-week high this week, 15 a new low. Global Tech: the week on our signal board. The board's read on 16 October: 12 names at new 52-week highs this week, 15 at new lows. Over 5 sessions, 30 names climbed back above their 50-day line and 6 fell below it. Strength by index: in the Agentic Winners 40, 65% of names are above their 50-day line. In Rubin Build-Out, only 19%. The names that stood out: Asia Vital Components +7.5%, Lasertec +5% — all at new highs. On the other side: Dell Technologies fell 4.6% on the day. Every card, every session, in the archive — and five views: movers, trends, overbought and oversold, drawdowns, analyst moves. The board, live every session: closelook.net/signals/board
- **Sectors — Energy and Staples Gain, Tech Slows** (The Board, 2026-10-17) → /videos/2026-10-16-the-board/
  Energy, Staples and Utilities gained the most ground on the sector rotation map, Tech and Communication lost the most. US sectors this week: staples +5.0% led, tech −7.7% trailed; 7 of 11 rose. Narration: AI voice.
  Transcript: Where the money moved this week, sector by sector. The Board: the week in sectors, regions and signals. The rotation map: right means stronger than the S&P 500 over a year, up means stronger over the last month. Energy, Staples and Utilities gained the most ground this week; Tech and Communication lost the most. US sectors this week: staples led with +5%, tech trailed with −7.7%. 7 of 11 sectors rose. Around the world: best Europe energy +5.8%, worst US tech −7.7%. On our signal board, the names that stood out: Asia Vital Components: up 7.5% at the close. Veeva: -0.9% at the close.
- **Market Breadth — The Leaders Are Tiring** (Divergence Watch, 2026-10-17) → /videos/2026-10-16-divergence-watch/
  15 of 26 divergences are active, 15 pointing down — risk score 54 of 100 (leadership exhaustion). Strongest: Small-cap. In the last month the S&P 500 rose 0.3%, the small caps fell 5.3%. Narration: AI voice.
  Transcript: 15 of 26 warning lights are on. 15 point down, 0 point up. Divergence Watch: where the market’s surface and its foundation disagree. Our scanner compares 26 pairs of markets that usually move together. Risk score this week: 54 of 100. The reading: leadership exhaustion. Strongest: Small-cap — 75 of 100, new today. Next: Small-cap upside-beta failure (87) and Breadth-momentum (85). In the last month the S&P 500 rose 0.3%, the small caps fell 5.3%. The S&P 500 made its high on 6 October. The small caps are still 11.4% below their August high. If it resolves up: Risk appetite broadens down the cap scale. If it resolves down: Large caps follow small caps lower.
- **AI Trends — Winners Still Lead, Rubin Fades** (Closelook@Charts, 2026-10-17) → /videos/2026-10-16-directional-flow/
  Directional Flow, the week to 16 October: trend strength rose most in the Agentic Winners 40 and fell most in Rubin. Among the big names it rose most in Palantir (+30), Salesforce (+25) and SAP (+25) and fell most in Micron (−24), Intel (−18) and ASML (−9); across the whole scan, heating up fastest were SABESP, Atlassian and INNIO, cooling fastest Xanadu Quantum Technologies, Zhipu AI and AXIA Energia. Narration: AI voice.
  Transcript: Trend strength this month: the Agentic Winners 40 up, Rubin down. Directional Flow: how strong a trend is, and how fast that strength changes — the change often comes before the price turns. Over the past month, trend strength rose most in the Agentic Winners 40 (+16) and the Agentic Ecosystem (+7). Rubin fell most (−17). The big names: trend strength rose most in Palantir (+30), Salesforce (+25) and SAP (+25). It fell most in Micron (−24), Intel (−18) and ASML (−9). Across all ~1,200 markets and stocks, heating up fastest: SABESP, Atlassian and INNIO. Also building: Caris Life Sciences and Zscaler. Cooling fastest: Xanadu Quantum Technologies, Zhipu AI and AXIA Energia. Also fading: Liquidia Technologies and Kioxia. Still rising, but losing strength: Liquidia Technologies, Kioxia and Applied Optoelectronics. Turning up from weakness: INNIO, monday.com and Bitcoin ETF. The full scan after every US close: closelook.net/signals/directional-flow
- **AI Trade — Crash, or Just a Pause?** (, 2026-10-15) → /videos/2026-10-15-ai-crash-or-pause/
  Korea fell more than 8% and the Nasdaq sits right on support — is the AI trade about to crash? Chip stocks never confirmed the new high. But Anthropic just made $16.8 billion in one quarter. The test comes with Microsoft on 28 October and Datadog and Cloudflare in early November: if they deliver, this was a pause; if not, AI spending pauses in 2027 — not a crash. Narration: AI voice.
  Transcript: Is the AI trade about to crash? In under a minute: the chart, the warning sign, and the date that decides it. Korea fell more than 8% overnight. The Nasdaq sits right on support. If that line breaks, the next stop is about 3% lower. Warning sign: chip stocks never made a new high with the Nasdaq. The big tech names carried it alone. The fear: 2000 all over again — huge spending, too little use. But Anthropic just made $16.8 billion in one quarter. The test: Microsoft on 28 October. Then Datadog and Cloudflare in early November. If they deliver, this was just a pause. If not, AI spending slows in 2027. A pause, not a crash. What to watch: the Nasdaq at 704, and Microsoft on 28 October. Follow for the update right after the report.
- **AI Trade — Show-Me Quarter for 2027** (, 2026-10-15) → /videos/2026-10-15-ai-trade-show-me-quarter/
  Two clocks on the AI trade: Korea −8.5% and Barclays' doubts on company AI spending against TSMC's record and ASML sold out into 2028. Companies rent AI rather than buy servers — Anthropic earned $1.70 per dollar of compute. The earnings of the next three weeks decide 2027. Narration: AI voice.
  Transcript: Korea −8.5%. TSMC: a record quarter. AI Trade: a show-me quarter decides 2027. On Thomas’s chart the Nasdaq-100 sits on support at 729, where the April trendline meets. If it breaks: 704 next, then the summer low near 662. The chip funds did not confirm the Nasdaq’s new high: SOXX, SMH, SMHX and XSD all stayed below June. A broad non-confirmation by the semiconductor complex. The biggest names carry the index: TOPT, the top-20 fund, broke above its summer high at 35. Our AI indices split: the Rubin build-out sits well below its June peak, the Agentic Ecosystem near its high. The short clock says sell: the chip fund is 20.7% below its June high. The long clock says intact: TSMC profit +68.7%, ASML sold out into 2028. The fear is 2000’s empty fibre. Barclays: firms put about 5% of IT budgets into AI. But companies rent AI. Anthropic: $16.8 billion this quarter, $1.7 per compute dollar. Our demand gauge reads 74.1, above the build-out at 56.5. The reverse of 2000. The watch item: two demand pulses cooled this week. The receipts: Microsoft on 28 October, Datadog and Cloudflare in early November. If they deliver, 2027 is intact. If not, expect a pause of six to twelve months.
- **Nvidia — Hopper, Blackwell, Now Rubin** (Closelook@Charts, 2026-10-11) → /videos/2026-10-09-chart-nvda-long/
  The long read on Nvidia, drawn by hand on our stock page: three chip generations in five years — Hopper in full production in September 2022 and the stock more than tripled in a year, the Blackwell year 2024 at +171%, Vera Rubin in full production since January. The steep line from the 2022 low broke on 27 February 2025; a wider channel took over and caught the lows of April 2025 and July 2026. The record of 239.24 on 6 October, where the desk stands, and what would change the read. Narration: AI voice.
  Transcript: Nvidia closed at a record on 6 October: 239. By Friday it was 4% lower. Closelook@Charts, the long read: Nvidia. In its last fiscal year, to January, revenue grew 65%, and 45% of it became free cash flow. And growth sped up: in the quarter to 31 July, revenue more than doubled, up 106%. Zoom out to five years. In 2022, Nvidia lost two thirds of its value; the low came in October, near 11 dollars. That September, its Hopper chip had gone into full production. A year later, the stock had more than tripled. 2024 was the Blackwell year: unveiled in March, shipping by the end of the year. The stock rose 171%. The steep line from the 2022 low held until 27 February 2025, the day after Blackwell’s first big quarter. Then it broke. A wider channel took over. Its lower line caught the lows of April 2025 and July 2026. In January, Vera Rubin went into full production. The desk’s open question: does Rubin move it up again? Now the past twelve months, day by day. From the October 2025 high, every rally stopped at one falling line, until it broke on 14 April. Ten days later it was back above its old record, at 207. A summer dip took it to 190 on 29 July, right onto the wide channel’s lower line. On 6 October it closed at a new record, 239. By Friday it was at 229. The desk added Nvidia on 29 June, during the dip, and again on 2 October. It sits in our wikifolio Core and Conviction at 9.6%, and in AI Cycle 2030 at 2.2%. The other two can’t hold it: Global Generation is ETFs only, The Compound is growth without tech. What would change the read? Above 239, record territory again. Below the wide channel’s lower line, now near 211, the channel would break. Next test: results on 18 November. Nvidia beat profit forecasts in all of its last 10 reports, and the shares fell after 6 of them. The full chart, the drawings and the record: closelook.net/indices/stock/nvda
- **Datadog — A Record Despite the AI Scare** (Closelook@Charts, 2026-10-11) → /videos/2026-10-09-chart-ddog/
  Datadog closed at a record on 9 October, 293.26, up 186% from its 23 February low — when investors feared AI agents from Anthropic and others could take over part of its job. Its rival Dynatrace hit a 52-week high the same day. The falling line, the 5 March break, the rising channel, the August record at 288, where the desk stands and what would change the read. Narration: AI voice.
  Transcript: Datadog closed at a record on 9 October. Closelook@Charts: one stock, one chart, one read. One year of Datadog. The low came on 23 February, at 103. Investors feared AI agents from Anthropic and others would take over part of its job. On 5 March it broke above its falling line. Since then: up 186% in 33 weeks, inside a rising channel. On 9 October it broke through the August record at 288. Its rival Dynatrace hit a 52-week high the same day. The desk added Datadog in Hypergrowth in May, and trimmed it in AI Buildout. It sits in our wikifolio AI Cycle 2030 at 2%. Back below 288, it would be under the August record again. Next test: results on 5 November. The full chart and the record: closelook.net/indices/stock/ddog
- **Palantir — The Chart Behind the Record Close** (Closelook@Charts, 2026-10-11) → /videos/2026-10-09-chart-pltr-long/
  The long read on Palantir: the handover from AI capex to opex, a 2025 Rule-of-40 score of 103, the rising channel since 2023 whose lower line caught the 25 June low, the double bottom with a higher low, the 4 August breakout gap, the channel from the June low and the record close of 209.05. Where the desk stands, and what would change the read.
  Transcript: Palantir closed at a record on 9 October: 209.05. This is the chart behind it. Closelook@Charts, the long read: Palantir in two and a half minutes. Palantir is part of the handover in the AI trade — from building AI to running it: from capex to opex. In 2025 it was one of the strongest software names: revenue up 56%, free cash flow 47% of revenue. Growth plus margin gives a Rule-of-40 score of 103 — more than twice the bar. This year growth sped up again: revenue rose 93% in the quarter to 30 June. Zoom out to 2023. On a log chart, Palantir has climbed inside one rising channel. The desk's lower line runs along the lows of May 2023 at 7.28 and August 2024 at 21.23. In 2026 the stock went sideways for months — and stopped on 25 June at 106, just above 100 dollars, on the same line. The upper line runs through the August 2025 high. The lower line rises about 136% a year. While the price stays above it, the long channel holds. Now the past twelve months, day by day. From the December high, every rally stopped at one falling line — the one the desk drew. The low came on 25 June at 107.27. The retest on 30 July held higher, at 122.26: a double bottom with a higher low. On 4 August, the day after its results, Palantir gapped from 126 to 145 — above the falling line — and closed at 162.66, up 29%. Since the June low it has climbed inside a rising channel — the desk drew it from the low. Up 95% in 15 weeks. Overhead sat last winter's highs: 200 in December, 208 in November. On 9 October it closed at 209.05, above both — the highest since Palantir listed. The desk added Palantir on 28 and 29 May, before the low: AI Buildout, Hypergrowth and Derivatives. A strategic holding, held and not traded: 9.4% of our wikifolio Core and Conviction, 2.1% of AI Cycle 2030. The other two can't hold it: Global Generation is ETFs only, The Compound is growth without tech. What would change the read? Above 208, record territory. Back below 200, last year's range again. The long channel's lower line now runs near 142. Below it, the trend since 2023 breaks. Next test: results on 2 November. Palantir beat profit forecasts in all of its last 10 reports. The full chart, the drawings and the record: closelook.net/indices/stock/pltr
- **AI Sectors — Software Up, Chip Hardware Down** (Closelook@AI Indices, 2026-10-11) → /videos/2026-10-09-ai-sectors/
  The best and worst of 35 AI sectors in Rubin, the Agentic Ecosystem and the Agentic Winners 40 — the week, the month and 2026 so far. This week application leaders led (+7.1%) and advanced packaging & bonding trailed (−8.0%). The AI board ranks every single stock too — this week's best: PTC +35.0%, Zscaler +18.9%, Money Forward +14.0%.
  Transcript: This week in AI: software up, chip hardware down. 35 AI sectors in three indices: the best and the worst of the week, the month and the year. The week: Agentic Winners 40 +4.9%, Agentic Ecosystem +2.8%, Rubin Build-Out 100 −3.4%. This year: Rubin +104%, Ecosystem +81%, Winners −4%. Best sectors this week: application leaders +7.1%, agentic security +7.0% and operations & observability +6.6%. Worst: advanced packaging & bonding −8.0%, storage −7.4% and power semiconductors −7.3%. Over the month, agentic security +29.0% and identity, trust & governance +26.6% led. At the bottom: storage −10.4% and compute operators −7.1%. The year so far: storage +313% and substrates & interposers +229%. The laggards: endpoints −18.1% and control plane −6.4%. The AI board ranks every single stock too — all 204 in the three indices, not only the sectors. Best this week: PTC +35.0%, Zscaler +18.9%, Money Forward +14.0%. Scan the code for all of them. The AI board, every sector and every stock: closelook.net/indices/ai-board
- **AI Trends — Winners Build, Rubin Fades** (Closelook@Charts, 2026-10-11) → /videos/2026-10-09-directional-flow/
  Directional Flow, the week to 9 October: trend strength rose most in the Agentic Winners 40 and fell most in Rubin. Among the big names it rose most in Palantir (+28), Salesforce (+28) and SAP (+25) and fell most in Micron (−22), Intel (−15) and ASML (−11); across the whole scan, heating up fastest were Atlassian, SABESP and SpaceX, cooling fastest Zhipu AI, Xanadu Quantum Technologies and AXIA Energia.
  Transcript: Trend strength this month: the Agentic Winners 40 up, Rubin down. Directional Flow: how strong a trend is, and how fast that strength changes — the change often comes before the price turns. Over the past month, trend strength rose most in the Agentic Winners 40 (+17) and the Agentic Ecosystem (+9). Rubin fell most (−17) — though its trend is still the strongest of the group. The big names: trend strength rose most in Palantir (+28), Salesforce (+28) and SAP (+25). It fell most in Micron (−22), Intel (−15) and ASML (−11). Across all ~1,200 markets and stocks, heating up fastest: Atlassian, SABESP and SpaceX. Also building: Caris Life Sciences and Zscaler. Cooling fastest: Zhipu AI, Xanadu Quantum Technologies and AXIA Energia. Also fading: Applied Optoelectronics and Planet Labs. Still rising, but losing strength: Zhipu AI, Applied Optoelectronics and Kioxia. Turning up from weakness: SABESP, Zscaler and INNIO. The full scan after every US close: closelook.net/signals/directional-flow
- **Global Tech — Palantir and Datadog at New Highs** (The Board, 2026-10-11) → /videos/2026-10-09-global-tech/
  The week on our tech signal board to 9 October: 23 names at a new 52-week high and 13 at a new low, 52 back above their 50-day line, the share of names above it in each index, and the stocks that stood out — Palantir, Datadog, Sonova.
  Transcript: 23 tech names hit a 52-week high this week, 13 a new low. Global Tech: the week on our signal board, in under a minute. The board's read on 9 October: a split market: 130 names in an uptrend, 149 below their 50-day line, 10 at new highs. Over 5 sessions, 52 names climbed back above their 50-day line and 9 fell below it. Strength by index: in the Agentic Ecosystem, 74% of names are above their 50-day line. In HALO Growth, only 25%. The names that stood out: Palantir +5.2%, Datadog +7.1%, Sonova +3.6% — all at new highs. On the other side: Renesas Electronics fell 4.0% on the day. Every card, every session, in the archive — and five views: movers, trends, overbought and oversold, drawdowns, analyst moves. The board, live every session: closelook.net/signals/board
- **Palantir — From the June Low to a Record Close** (Closelook@Charts, 2026-10-11) → /videos/2026-10-09-chart-pltr/
  One year of Palantir in one chart: the falling line that capped every rally from December to June, the break on 4 August, the rising channel from the 25 June low (+95% in 15 weeks), last winter's highs at 199 and 207, and the record close of 209.05 on 9 October. Where the desk stands, and what would change the read.
  Transcript: Palantir closed at a record on 9 October: 209.05. Closelook@Charts: one stock, one chart, one read — from the stock page. One year of Palantir, day by day: the low on 25 June at 107.27, a record close on 9 October. From December to June, every rally stopped at one falling line. On 4 August, the day after its results, it jumped 29% and closed above that line. Since the June low, Palantir has climbed inside a rising channel. Up 95% in 15 weeks. Overhead sat last winter's highs: 199 in December, 207 in November. The close on 9 October, 209.05, cleared both — the highest since Palantir listed. The desk added Palantir on 28 and 29 May: AI Buildout, Hypergrowth and Derivatives. A strategic holding, not traded: 9.4% of our wikifolio Core and Conviction, 2.1% of AI Cycle 2030. The other two can't hold it: Global Generation is ETFs only, The Compound is growth without tech. Our chart on the stock page draws the same picture: the falling line, the channel, the 199–207 zone. What would change the read? Above 207, record territory. Back below 199, last year's range again. Next test: results on 2 November. Palantir beat profit forecasts in all of its last 10 reports. After those 10 reports the shares rose 5 times and fell 5 times. The full chart, the drawings and the record: closelook.net/indices/stock/pltr
- **Records Without the Chips — the Divergence Now** (Closelook@Charts, 2026-10-10) → /videos/2026-10-10-charts-divergence-now/
  The S&P 500 and the Nasdaq 100 set records on 6 October and cloud funds hit new highs — but the chip funds are 10–19% below their June highs. Meanwhile staples, discretionary and financials bounced from their late-September lows. Next week’s earnings may give the clue.
  Transcript: The S&P 500 and the Nasdaq 100 are at records. The chips are not. When the leaders stop confirming, the charts send a message. Here is today’s. The S&P 500 and the Nasdaq 100 set records on 6 October. Cloud funds hit new highs on Friday. Chips did not: SOXX is 15% below its June high, the equal-weight XSD 19%. Since 22 June the Nasdaq 100 is 2% higher. The chip fund is 15% lower. A new record without the old leaders: that is the divergence. Meanwhile the rally widened: staples, discretionary and financials bounced from their late-September lows. This week: consumer staples +3.6%, discretionary +2.6%, financials +2.3%. Next week’s earnings may give the clue: the big banks Tuesday, ASML Wednesday, TSMC Thursday. Do the chips catch up, or does the market move on without them?
- **Three Times the Charts Warned First in 2026** (Closelook@Charts, 2026-10-10) → /videos/2026-10-10-charts-divergences-2026/
  A divergence is when two charts that usually move together stop agreeing. In March Rubin held while the Nasdaq 100 made a new low — Rubin then gained 86% in the second quarter. In June the equal-weight chip fund failed to confirm the high. In spring software held its February low and cloud funds rose about 70%.
  Transcript: Charts often warn before prices move. Three times this year, they did. A divergence: two charts that usually move together stop agreeing. March: the Nasdaq 100 fell to a new low on 30 March, 6% below its 5 February level. Our AI build-out index Rubin did not. It held 4% above its February level. That was the tell. In the second quarter Rubin gained 86%, the Nasdaq 100 28%. June: the big chip fund SOXX made a new high on 22 June. The equal-weight chip fund XSD did not. It stopped just below its 3 June high. That was the top. Since then SOXX is down 15%, XSD 19%. Spring: software fell back to its February low in April — and held there. From the April low, cloud funds rose about 70%, software 51%, the Nasdaq 100 23%. Next: the divergence that is open right now.
- **HALO — Growth AI Can’t Touch** (Closelook@AI Indices, episode 5, 2026-10-10) → /videos/2026-10-10-ai-indices-ep5-halo/
  The other end of the barbell: 100 growth companies in 12 themes, from defense and longevity to the grid and space, whose growth does not depend on the AI race. Down 2% this year — but since the June chip high Rubin fell 15% and HALO only 5%.
  Transcript: The other end of the barbell: growth that AI can’t touch. Episode 5: HALO, the ballast of the AI trade. 100 companies in 12 themes: defense, longevity, the grid, nuclear, space, payments and more. Their growth does not depend on who wins the AI race. This year HALO is down 2%. It peaked in late January and sits 12% below that high. Space +21%, grid and materials +16%, longevity +11%. Clean energy −29%, travel and leisure −24%. Since the June chip high, Rubin fell 15%. HALO fell 5%. The ballast did its job: it fell less when the builders fell. Next, a new format: what the charts say.
- **Agentic Winners 40 — Who Gets Paid** (Closelook@AI Indices, episode 4, 2026-10-10) → /videos/2026-10-10-ai-indices-ep4-agentic-winners/
  Forty companies in four tiers that have to show they gain from AI agents. Down 4% this year but up 36% since the 25 June low — since the June chip high the application leaders are up 50% and the control plane 41%.
  Transcript: The builders got paid first. This index tracks who gets paid next. Episode 4: Agentic Winners 40 — the applications. 40 companies in four tiers: megacap gateways, control plane, application leaders and endpoints. Each name has to show it gains from AI agents, not just talk about them. This year the index is still down 4%. But since the 25 June low it is up 36%. This year only the megacap gateways are clearly up, +18%. The endpoints, the devices, are down 18%. Since the June chip high the order flipped: application leaders +50%, control plane +41%. The money moved from the chips to the software that uses them. Next: HALO — growth AI can’t touch.
- **Agentic Ecosystem — The Rails That Run AI** (Closelook@AI Indices, episode 3, 2026-10-10) → /videos/2026-10-10-ai-indices-ep3-agentic-ecosystem/
  Rubin builds the AI factory; these 34 companies run it and get paid every time an AI agent runs. Up 81% this year with a record close on 9 October — since the June chip high the govern-and-secure layer is up 58% while Rubin fell 15%.
  Transcript: Rubin builds the AI factory. This index runs it. Episode 3: the Agentic Ecosystem — the operating cost of AI. 34 companies: compute renters, data and memory, runtime, monitoring, identity and security. They get paid every time an AI agent runs, not when a chip is built. This year the index is up 81%. On Friday it closed at a record, almost double its February low. Since the June chip high, govern and secure +58%, execution +44%. The substrate layer, closest to the hardware, fell 8%. Rubin fell 15%. Last month security led: agentic security +29%, identity and trust +27%. The compute renters fell 7%. Next: Agentic Winners 40 — who gets paid.
- **Rubin — Inside the AI Factory Build-Out** (Closelook@AI Indices, episode 2, 2026-10-10) → /videos/2026-10-10-ai-indices-ep2-rubin/
  Rubin holds 130 companies in eight layers, from chip design software to robots, all equally weighted. It is up 104% this year but about 15% below its 22 June high — and in the last month the leadership moved from storage and memory to chip design software and interconnects.
  Transcript: Rubin tracks the companies that build the AI factory. Episode 2 of our series on the AI trade: one index, up close. 130 companies in eight layers, from the software that designs chips to the robots that use them. Every name has the same weight, so a small supplier counts as much as Nvidia. This year Rubin more than doubled: +104% since 2 January. It peaked on 22 June and sits about 15% below that high. The bottlenecks paid most: connectivity and materials +153%, assembly and test +130%. Manufacturing support rose least, +37%. In the last month the leaders changed: chip design software +19%, interconnects +13%. The year’s biggest winners fell: storage −10%, memory −2%. Next: the Agentic Ecosystem — the rails that run AI.
- **Sectors — Health care and Staples Gain, Tech Slows** (The Board, 2026-10-10) → /videos/2026-10-09-the-board/
  Health care, Staples and Materials gained the most ground on the sector rotation map, Tech and Communication lost the most. US sectors this week: utilities +4.0% led, tech −0.5% trailed; 9 of 11 rose. On the signal board: 130 stocks in an uptrend, 10 at new highs.
  Transcript: Where the money moved this week, sector by sector. The Board: sectors, regions and signals in under a minute. The rotation map: right means stronger than the S&P 500 over a year, up means stronger over the last month. Health care, Staples and Materials gained the most ground this week; Tech and Communication lost the most. US sectors this week: utilities led with +4.0%, tech trailed with −0.5%. 9 of 11 sectors rose. Around the world: best Europe tech +4.0%, worst Europe communication −2.9%. On our signal board: 130 stocks in an uptrend, 149 below their 50-day line, 10 at new highs. Palantir: new 52-week high. Agentic Ecosystem: at a record high.
- **Market Breadth — Records on a Thin Base** (Divergence Watch, 2026-10-10) → /videos/2026-10-09-divergence-watch/
  20 of 26 divergences are active, 20 pointing down — risk score 64 of 100 (mega-cap concentration warning). Strongest: Nasdaq breadth. In the last month the Nasdaq 100 rose 6.0%, the equal-weight Nasdaq rose 3.7%.
  Transcript: 20 of 26 warning lights are on. 20 point down, 0 point up. Divergence Watch: where the market’s surface and its foundation disagree. Our scanner compares 26 pairs of markets that usually move together. Risk score this week: 64 of 100. The reading: mega-cap concentration warning. Strongest: Nasdaq breadth — 85 of 100, 15 days running. Next: Small-cap (82) and Breadth-momentum (82). In the last month the Nasdaq 100 rose 6.0%, the equal-weight Nasdaq rose 3.7%. The Nasdaq 100 made its high on 6 October. The equal-weight Nasdaq is still 2.2% below its August high. If it resolves up: Nasdaq participation broadens. If it resolves down: Nasdaq 100 fades after equal-weight deteriorated.
- **The Barbell — Four Sides of the AI Trade** (Closelook@AI Indices, episode 1, 2026-10-09) → /videos/2026-10-09-ai-barbell-ep1/
  Four indices, one picture of the AI trade: Rubin tracks what builds it, the Agentic Ecosystem what operates it, Agentic Winners 40 who captures the value, and HALO what AI can’t touch. The builders ran furthest this year — since the June chip high the order has flipped.
  Transcript: Four indices. One thesis. The whole AI trade in one picture. Episode 1 of a short series on how we track the AI trade. Rubin tracks what builds the AI factory: chips, memory, power and cloud. Agentic Ecosystem tracks what operates it: the rails that run, secure and bill AI agents. Agentic Winners 40 tracks who captures the value. HALO tracks what AI can’t touch: durable growth outside the AI build-out. This year the builders ran furthest: Rubin +103%, Agentic Ecosystem +74%. Winners 40 and HALO are still below where they started the year. Since the chip high in June the order has flipped: Winners +30%, Ecosystem +21%, Rubin −16%. That is the barbell: three sides of AI on one end, what AI can’t touch on the other. Next: one index at a time, starting with Rubin.
- **Capex Sold, Apps Bought** (Closelook@AI Trade, 2026-10-09) → /videos/2026-10-09-ai-trade-capex-sold-apps-bought/
  OpenAI’s revenue came in at $50 billion, not the $70 billion that had circulated. Chips fell and the software that uses AI rose — since the 22 June chip high, build is down, operate and use are up.
  Transcript: OpenAI’s revenue came in at $50 billion, not the $70 billion that had circulated. Four things in under a minute. Chips fell. The software that uses AI rose. Since the chip high on 22 June, our three AI indices went separate ways. Build −15.8%. Operate +21.1%. Use +29.8%. Across the AI chain, cloud and cybersecurity lead. Chips, data centers and power trail. In every tech cycle, the profits moved from those who build to those who use. No reason to leave tech. A reason to watch where the money goes next.
<!-- auto:videos:end -->

## Author & publisher (E-E-A-T)
- **Thomas Look** — Managing Director, Closelook Venture GmbH; background in trading, quantitative research, neural networks, and technology businesses → /about/thomas-look/ , /about/
  - LinkedIn: https://www.linkedin.com/in/thomas-look/
  - X/Twitter: https://x.com/ThomasLook5
  - Substack: https://closelooknet.substack.com
- Editorial policy & methodology → /about/editorial-policy/

## Contact
- Web: https://closelook.net
- Newsletter / paid access: https://closelooknet.substack.com

## Full page catalog
- Machine-generated catalog of every indexable page (title + description + URL), regenerated at each build: https://closelook.net/llms-catalog.txt