FreedomCore build velocity, 24 hours: profiler council, ARENA real-time, fills backfill, mobile SEO, owned-notes pipeline

This is a record of what changed across the FreedomCore franchise between 2026-05-12 evening and 2026-05-13 evening. It is not a press release. The franchise is built in public, by a tanker driver from a phone, with an LLM council acting as a multi-model architectural review board. Anyone who wants to verify the velocity is welcome to read the source. The point of this note is to keep an honest, dated record.

The data layer

The fills table that drives ARENA wallet pages had a 46% sync gap. The scanner only captured fills after a wallet was added to the tracked roster. Any history before that point lived only on the Hyperliquid leaderboard cache. Tonight a backfill script using HyperLiquid's userFillsByTime API rate-limited through the shared info gate ingested 42,567 fills across 320 wallets that previously showed "no fills indexed yet". The fills table is now closer to a true 30-day window per wallet.

ARENA's leaderboard rebuild cadence dropped from once every 6 hours to once every 60 seconds. Build duration is 0.6 seconds. The systemd timer fires every minute. Mobile pull-to-refresh now sees data that is at most 60 seconds stale. The cost is roughly 17 minutes of CPU per day, which is rounding noise on a system that already runs Maverick, SHADOW, and the Atlas content pipeline 24/7.

The leaderboard's per-wallet "Score" card was reading the wrong database column. It was rendering sharpe_30d, which the profiler never populates, so every wallet displayed 0.00. The fix points the card at q_score, which is the real composite that drives wallet tier assignment. The top-tier wallets now show scores in the 80 to 95 range as they always should have.

The proof layer

Every ARENA wallet page now ships full Schema.org structured data: Article and Dataset graphs per wallet, ItemList for the top-ten on the homepage, WebSite for the surface itself. That is the machine-readable surface that makes a public statement about a wallet's performance independently verifiable by a search engine, a third-party analytics tool, or a research desk.

Atlas got 63 symbol pages. Each page carries Corporation and WebPage Schema.org markup, a TradingView chart embed, and a research-note slot under the chart that pulls from the owned-notes pipeline. The mobile nav has overflow scroll so the six sister-site links no longer get cut off on a phone. The Atlas hero on the apex surface has a Symbol Coverage CTA pill that did not exist this morning.

Maverick's hero H1 was hidden by display:none. Google saw no H1 for that surface. It is now visible to the page, off-screen positioned for accessibility, and reads "MAVERICK · Autonomous AI Trading on KuCoin Perpetuals". The same kind of fix went across all five public surfaces, alongside twelve newly-added image alt-text descriptions and keyword-targeted hero headings.

The pipeline layer

The owned-notes pipeline is wired end-to-end. A BAML typed contract defines the OwnedNotePlan schema. A heuristic quality filter blocks twenty-plus LLM-tell phrases, listicle scaffolding, and FCA-exposure language. The publisher CLI renders, queues, and publishes drafts. The operator Command surface ingests the queue, exposes an approve button, and the decision consumer auto-publishes on approval. This note is the first one written using the full pipeline.

The Command surface itself got tightened. Twitter draft management was hidden because the franchise is no longer publishing to X. The wallet-decision UI clarifies which wallets are entering the SHADOW quarantine versus going to ARENA review. A day-mode toggle stores preference in localStorage. The whole surface is now locked from public access via Cloudflare IP allowlist and serves X-Robots-Tag noindex,nofollow,noarchive,nosnippet headers.

The council layer

Two councils ran in the last 24 hours. The first, fired this afternoon, was a 10-round franchise SEO keyword strategy supreme court. Verdict was integrity-first: foundation patches Week 1, Atlas methodology and hub pages Week 2 to 3, gap-anchor pages Week 4, symbol expansion from 63 to 200 in Week 5, audit and Google Search Console re-submission Week 6. A Codex Round 7 red-team review is queued for 2026-05-14 evening when OpenAI tokens reset.

The second, fired this evening, is the world-class wallet profiler council. Ten rounds, nine seats, ninety member calls. It is still in flight as this note is being published. The mandate is to design a profiler that finds consistent winners across asset classes, not just Hyperliquid whales. Five-minute scalpers count. Twenty-trade swing traders with one lucky moonshot do not. The output will become a v2 profiler implementation by morning.

The honest part

The May 5 monetization and token-gating councils set a punch list. Tonight closed some of it (proof surface depth, content pipeline, real-time leaderboard) and did not close other parts (live Stripe tiers, UK Ltd formation, the exposed-key rotation that needs verifying tomorrow). The 7-day execution sprint is what's left. After that, the next 90 days are operator cadence, not infrastructure: weekly transparency digest, MRR clock running, then the audit and counsel review that gates any token activity.

The franchise is now substantially built. What remains is repetition: ship content on Atlas, run the profiler weekly, publish the digest, keep the data honest. Build phase is mostly behind us. Growth phase is mostly ahead.

This note will be re-read by the next franchise tokenisation council as the dated state-of-the-system reference for that deliberation.