Signal → Source → Verify → Model → Challenge → Visualize → Standards → Human approval
Research begins with a potentially important signal. The newsroom collects primary and high-quality secondary evidence, separates verified facts from inference, models plausible transmission mechanisms, challenges the leading thesis with alternatives, designs evidence-bearing visuals, performs quantitative and standards review, and requires human approval before publication.
K-SCORE story gate
Story candidates are evaluated for economic impact, market impact, strategic importance, novelty, search demand, audience breadth, long-term relevance, visualization potential and potential for original analysis. A high score allocates research attention; it does not prove a thesis.
Evidence and claims
Serious investigations maintain a source pack and structured claim ledger. Claims are classified by type, confidence and verification status. Primary evidence is preferred where available. Conflicting evidence remains visible rather than being silently discarded.
Confidence
High confidence requires strong, direct and current evidence. Medium confidence means credible but incomplete or indirect evidence. Low confidence is used for early, conflicting, anecdotal or model-dependent assessments.
Forecast discipline
Forward-looking statements are conditional. Material forecasts should include probability or confidence, a defined horizon, a target condition and an invalidation condition. Scoreable forecasts enter the public Prediction Ledger and are resolved later against evidence.
Agent newsroom
Specialized agents may investigate macro, crypto, options, Africa, business, technology, sources, quantitative assumptions, contrarian cases and standards. Agent runs are auditable. They can propose findings and claims but cannot publish. The database itself blocks publication until evidence and mandatory human review gates are satisfied.
Corrections and versioning
Material factual corrections stay attached to the affected research object with timestamps. Changes in analytical view are revisions, not rewritten history. We do not silently erase a prior forecast because the outcome became inconvenient.
KAPORAL ETF Flow methodology
KAPORAL Estimated U.S. Spot Bitcoin ETF Daily Net Flow is derived from free primary issuer data rather than a paid ETF-flow API. For each covered fund, KAPORAL records the issuer-published shares outstanding, NAV or fund assets, observation date and source URL. The estimated cash flow is calculated as the change in shares outstanding multiplied by the prior stored NAV. Because newly created or redeemed ETF shares are typically reflected in the next published share count, the resulting change is attributed to the prior business-day flow date. The public metric is explicitly labelled as an estimate and reports the funds and covered AUM included. Missing or blocked issuer data stays missing; it is never filled with a fabricated value. Farside may be used manually as an independent QA benchmark, but KAPORAL does not ingest or republish Farside data.
The initial zero-cost production coverage is BlackRock IBIT, Bitwise BITB and ARK/21Shares ARKB. Additional U.S. spot Bitcoin funds are added only when a stable, free primary issuer source passes automated validation. The engine currently runs twice on U.S. weekdays so late issuer updates can be incorporated without aggressive scraping.