Forecasts scored against outcome.

Current tally

Tracking active forecasts. Each one is logged at publication and scored as the outcome window closes.

The examples below are curated: first example of each category: structural, statistical, methodological, regime-specific, canon. Not a statistical sample.

Forecast accuracy

Each row is a forecast compared against the BTC buy-and-hold baseline. The out-of-sample window opened recently.

Backtest summary

Methodology changelog

Index updates and constituent changes are published here in standard changelog format.

1. First canonical rebalance event

   2026-05-17

   The first canonical rebalance event records the moment at which MAJOR Top-50 and BROAD Top-500 universes lock their methodology version and constituent membership. The event captures point-in-time universe snapshots: 50 constituents for MAJOR Top-50, 391 of a 500 target for BROAD Top-500. The methodology version, canonicalization version, and authority signature are frozen at the event timestamp. Subsequent reconstitutions follow a quarterly cadence and produce successor events that link back to this first one. This event closes the gap where index methodology specifications were published while the homepage still read from the unbranded broader universe; production routing of homepage and daily archive to the named tiers is the next ship cycle.

   scope

2. Index family scope clarified

   2026-05-15

   MAJOR Top-50 and BROAD Top-500 methodology specifications are published, each with documented constituent, cohort, and review-buffer parameters. The operating universe is still the broad family; migration to the named tiers is in progress. BROAD Top-1500 and Specialty Memecoin-100 are in development.

   scope

3. Observation axes treated as discovery outputs

   2026-05-15

   Per-substrate diagnostic axes (axis count, sector count, orbit-class assignment) are now declared as observation outputs rather than baked-in inputs. New substrates run a discovery pass before the methodology is parameterized for that substrate. This matches the convention used in academic index families: methodology evolves with substrate maturity and changes are versioned.

   methodology

4. Audit chain reliability upgrade

   2026-05-15

   An audit incident registry now tracks cases where the published methodology audit emitted a clean status before the mechanical verification chain ran to ground truth. The first entry documents an instance where the audit ran amber across roughly eighteen consecutive commits before the upstream blocker was found and addressed in a single fix. Prevention mechanisms (local commit gate, handoff verification block, expanded check scope) are in development.

   audit

5. Out-of-sample reference window opened

   2026-04-22

   BTC Season and Alt Season forecast accuracy scoring window opened 2026-04-22 through 2026-05-22. Out-of-sample entries accumulate over this window.

   scoring

6. Reference universe broadened

   2026-03-15

   On-chain (Hyperliquid) reference universe broadened to 1,300+ tokens. Prior 165-token universe deprecated as not structurally representative.

   universe