Atlas research notebook: 224 symbol pages, 56 indexed, and the SPY metadata canary now running

The Atlas research notebook now publishes 224 symbol pages with 176 indexable and 56 of them indexed in Google Search Console, while a single controlled SPY metadata canary runs toward a 2026-06-21 readout.

Coverage state on 2026-06-15

The public source-of-truth dashboard, read on 2026-06-14, records 224 symbol pages: 176 indexable, 48 noindex, and 2 deprecated redirects, organised under 6 asset-class hubs. Sitemap coverage lists 176 symbols, and the stocks sitemap at stocks/sitemap.xml carries 183 URLs. Daily contexts validate at 226, with the Atlas Pulse JSON validating 224 of 224 and 0 ETF hard-gate violations. The writer queue sits at 115 done, 0 failed, 0 pending, and 0 running, which means no half-written page is exposed to a crawler.

The daily rebuild ran 2026-06-15 at 05:09 UTC. Market-universe lists otherlisted.txt and nasdaqlisted.txt refreshed at 05:13 UTC, and the daily batch cache atlas_daily_batch_20260615_050102.json was written at 05:09 UTC. These timestamps are the spine: a research note is only as trustworthy as the dated rebuild behind it, so each context carries a provenance trail rather than a freshness claim asserted in prose.

What Search Console actually shows

The GSC coverage export, charted to row 2026-06-08, reads 56 indexed and 67 not indexed against 17 impressions. That 56 is the end of a measured climb: 12 indexed on 2026-05-29, rising to 56 by 2026-06-06. The not-indexed bucket breaks down into 13 deliberate noindex, 7 alternative canonical, 29 discovered not indexed, and 18 crawled not indexed, so most of the gap is either intended or a normal crawl-budget queue rather than a defect.

The 2026-06-14 performance export logged 47 page-report impressions, with a later same-day read showing 68, and the homepage moved from 0 to 1 click. Top symbol pages by impressions were /stocks/SPY/ at 14, /stocks/COIN/ at 10, /stocks/JPM/ at 9, and /stocks/PLTR/ at 3, with one homepage impression at position 9. Top queries included "coinbase global inc", "jpm stock", "pltr", "spy etf", and "jp morgan stock", which maps cleanly onto the highest-impression pages. You can read the published pages behind those figures through the Atlas symbol coverage index.

The SPY metadata canary as cross asset market intelligence test

A SEO canary went live 2026-06-14 at 19:54 BST under experiment id atlas_spy_title_meta_canary_20260614. It changes static metadata on the single SPY page only: a new title, "SPY ETF Analysis, Chart & Daily Brief | FreedomCore Atlas", plus matching meta, OG, and Twitter tags. Body copy, JSON-LD, canonical, and robots are unchanged. Hold-out controls are META, ORCL, GOOGL, MSFT, and SQ. The cooldown date is 2026-07-05, with a readout target near 2026-06-21. The page_state.json and the stocks/sitemap.xml lastmod entries for /stocks/, /stocks/etf/, and /stocks/SPY/ were set truthfully to 2026-06-14, so the change date in the sitemap matches the actual edit.

This is deliberately narrow macro market structure notes discipline: one variable, named controls, a fixed cooldown, and a no-churn freeze on indexable pages until the readout lands. The reasoning is recorded in the dated Atlas notes thread rather than re-derived each session.

Source rails and the economic calendar research feed

The public homepage stays indexable at https://atlas.freedomcore.io/, while the heavy research workstation moved to noindex,follow at https://atlas.freedomcore.io/workstation/. Source-authority bands cite official rails as named text links: SEC EDGAR, FRED, Bank of England, ONS, Nasdaq, TradingView, and CoinGecko, with no third-party logos. The OpenBB DefiLlama RSS market spine that feeds these contexts draws on documented public tooling, including the OpenBB docs and on-chain aggregates from DefiLlama, with each context carrying a structured trail consistent with Schema.org BlogPosting.

The economic-calendar API returns provider_state values of official_public_schedule and OFFICIAL_SOURCE_READY, exposing FOMC, PCE, and GDP schedule rows with source provenance. Actual, forecast, and previous values are not inferred when a provider has not supplied them, which keeps an empty cell honest instead of fabricated. The public dataset catalogue, published 2026-06-13 with row_count_public 176 and row_count_total 226, serves symbol-coverage JSON, CSV (now text/csv), schema, and manifest, and /datasets/ returns 200.

Discovery authority versus third-party authority

The latest links export shows 290 referring rows: 273 from arena.freedomcore.io, 14 from shadow.freedomcore.io, 2 from freedomcore.io, 1 from maverick.freedomcore.io, and 1 from quantum.freedomcore.io. That is owned-franchise discovery authority routed across FreedomCore properties and adjacent surfaces such as Maverick, not yet independent third-party authority. Backlink acquisition runs in parallel with the freeze, and the next structural build is semantic "what changed" plus source-delta receipts on the noindex workstation before any second canary.

Caveats and what is not yet proven

The homepage move from 0 clicks to 1 click is early and noisy and should not be read as a trend. The SPY canary verdict is not in until roughly 2026-06-21, and SPY sits around search position 66, so its click signal may be inconclusive at readout. Every referring domain in the link graph is still owned-franchise rather than an independent third party, so external authority remains unproven. The 29 discovered-not-indexed and 18 crawled-not-indexed pages remain outstanding, and the coverage chart itself lags by several days, so the 56-indexed figure describes 2026-06-08 and not the live moment. None of these figures are a forecast; they are dated readings of a public surface that the next crawl will revise.