See the direction, the evidence coverage, and what could invalidate the read.

What the system evaluates

The theme model combines several evidence families. A single strong day is never treated as a complete trend.

• Core-company breadth: how much of the mapped basket is advancing or declining.
• Sector benchmarks: whether relevant ETFs confirm the same direction.
• Time horizons: separate coverage for 1D, 5D, and 20D price evidence.
• Company news: recent catalysts from the highest-impact mapped names.
• Relevant events: prediction-market changes used as timing context, not the final conclusion.

Evidence gates before confirmation

A read is labeled confirmed, developing, or insufficient. The label depends on actual horizon coverage and data quality rather than on how dramatic the latest move looks.

• One-day-only evidence remains developing and waits for 5D confirmation.
• Missing horizons are not reweighted as if they were present.
• Unusually large moves trigger a split, adjustment, or source-data warning.
• Low basket coverage prevents a directional call.

How to read the result

Start with the direction and evidence-stage label, then inspect the three strongest drivers and the invalidation risks. Key assets come next; the full basket stays available in a disclosure rather than overwhelming the first screen.

Evidence coverage describes how complete the observation set is. It is not a probability of profit, a price target, or a trade recommendation.

What AheadPick publishes—and protects

Theme reports disclose the evidence needed to evaluate an output: basket coverage, horizons, benchmarks, news context, timestamps, warnings, and invalidation conditions.

Exact feature engineering, weights, thresholds, theme mappings, event-to-asset mappings, and internal prompts remain proprietary. Data-source disclosures are listed in the legal pages.