Screens and how their data is produced

Ask the AI Console in plain language to screen the stock universe by the criteria you care about, and it runs a multi-filter screen and renders the matches as a sortable table in the conversation. This page explains the screen-result table, how you describe a screen, how to read the columns, and the standard math behind the one derived figure. Screening runs inside the AI Console and is available on the Free, Pro Lite, and Pro plans, with each screen drawing down GNG credits like any answer. The proprietary GNG-score columns appear for paid tiers and are held back on Free.

Before you begin

Sign in to your account. Screening runs inside the AI Console.
For the meaning of each rating and valuation column, see the Research Terminal overview.

What it shows

A stock-screener capability inside the chat. You describe the screen conversationally, for example "large-cap dividend growers under a 20 price-to-earnings", and the console runs it and renders the results inline as a table widget.
One row per matching company, with the columns the screen returned. By default that is symbol, name, sector, asset type, recent close price, market capitalization, price-to-earnings, dividend yield, and 5-year dividend-per-share growth. Paid tiers also receive the proprietary GNG scores (overall GNG rating, GNG Quant, Wall Street rating, analyst target, Safety score, and Quality score).
A capped result set, so the table stays readable. The screen returns a bounded number of best-matching rows rather than the entire universe.
A header you can use to open the same screen in the Research Terminal with the filters pre-loaded.

Inputs

The screen is driven through conversation, and underneath it accepts the inputs in the following list. You express these in natural language and the console translates them.

Filters: one or more criteria, each a column plus a comparison plus a value, for example sector equals a value, market cap greater than a value, dividend yield greater than a value, or price-to-earnings less than a value. Comparisons include greater-than-or-equal, less-than-or-equal, equal, not equal, is-in-a-list, between two values, is-not-empty, and text-contains.
Sort: an optional column and direction. The default order is by symbol ascending when you do not specify one.
Return columns: an optional list of which columns to return. The default is the standard column set above. On Free, the proprietary GNG-score columns are stripped from whatever you ask for.
Result limit: how many rows to return. The default is 25, and the screen caps the maximum so the output stays manageable.

Outputs and how to read them

A ranked table of companies that satisfy all your filters, sorted as requested.
Standard columns (price, market cap, price-to-earnings, dividend yield, dividend growth, sector) are plain, transparent fundamentals you can read directly.
GNG proprietary columns (the GNG rating, GNG Quant, Wall Street rating, Safety, Quality, fair-value-derived fields, and buy-zone prices) are summary scores. Read them as GNG's at-a-glance verdict on valuation, safety, and quality on a defined scale, not as raw inputs. They appear for paid tiers only.
On Free, the table still returns the matching names and the standard public fundamentals; the proprietary columns are omitted, and an upgrade prompt explains how to see them.

The header states the filters in words and the total number of matches, which can exceed the rows shown because the row count is capped. Read the table as a starting shortlist, then open a name in the Research Terminal for the full picture.

How it is calculated

Filtering and sorting are standard, transparent operations on the dataset. The screen keeps only rows that satisfy every filter you gave, orders them by your chosen column, and returns the top rows up to the limit.

For example, a filter set of "market cap greater than 10 billion" and "dividend yield greater than 3 percent" and "price-to-earnings less than 18", sorted by dividend yield descending with a limit of 25, returns up to 25 companies that meet all three conditions, highest yield first.

Data and timing

Screens read the platform's consolidated research dataset, built from company fundamentals and pricing. The underlying inputs come from AlphaVantage (including Nasdaq-listed quotes delivered through AlphaVantage) and from EDGAR. The dataset is refreshed on the platform's regular update schedule, so a screen result reflects the most recent processed data rather than live intraday ticks, and it matches the Research Terminal at the same moment.

Use cases

Build an income-focused candidate list with dividend-yield and dividend-growth filters, without leaving the chat.
Find undervalued names in a sector by combining valuation filters with the GNG scores (paid tiers).
Narrow a large universe to a short, ranked shortlist, then ask the console to research each name.
Sanity-check an idea, for example whether many sub-20 price-to-earnings names exist in a sector, in seconds.

Limitations and disclosures

A screen presents ratings, valuations, and analyst figures. It is a research shortlist, not a recommendation. The total match count can exceed the rows shown because the row count is capped, so tighten the filters rather than assuming the visible rows are the whole result.

GNG Research provides equity research and educational tools, not investment advice. Nothing on the platform is a recommendation to buy or sell any security. Do your own research and consider your circumstances before making any investment decision.