Methodology and disclosures

How Kuber Market Intelligence prepares, checks, and scores its reports.

How reports are prepared

Prepared with local AI assistance and automated validation. Prices, dates, and counts come from stored source records and fixed calculations. The language model does not supply numbers, sources, or citations, and any sentence that adds a number the sources do not contain is discarded.

How stocks are chosen

A fixed screen (screen-v1) ranks names on news, liquidity, momentum, options activity, earnings, and event risk. When a factor has no data it is left out and the rest are re-weighted. A name must clear a minimum score, so some days list fewer than three stocks, or none. Reports never add names to fill the space.

How forecasts are scored

The morning list is frozen when it is published. The closing review scores every name against its reference price. A direction counts as supported only when the move reaches 0.5% in that direction. Watch-only names are never counted as directional wins. A missing or incomparable close is shown as not evaluable, not estimated. Results describe forecast direction and are not a trade, a fill, or an account return.

Data sources

Each report lists the sources it used. Quotes are delayed vendor readings with their source time in Eastern Time and are not exchange prints. Headlines are linked, not copied, and a headline is never presented as the cause of a market move. Missing data is left out instead of guessed.

Corrections

A correction is added as a visible update with its time. Earlier versions are kept, and a morning forecast is never edited after it is published.

Educational use only

This material is educational research. It is not individualized investment advice, a solicitation, a guarantee of results, or an account return. Markets can move against any view.