Methodology

How we rate stocks

Fixed rules, the same for every company, from official filings and market prices. No one sets a score by hand, and AI does not set one either.

Where the numbers come from

Fair value

We value each company several ways from its filings: the cash it makes, what its shares have usually cost, and what the market pays for similar businesses. Those are combined into one fair value, with a range and how sure we are of it.

A fair value is an estimate, not a price target. Research services often land 20% or more apart on the same company, and the app shows each method’s figure so you can see where ours comes from.

The PreOpen Score

Every company gets a score from 0 to 10, built from five parts:

40%
Where it ranks on value and quality together, among the S&P 500 companies we value
20%
Its price against its fair value
20%
The quality of the business: its returns, margins, growth and financial strength
10%
What its chart shows
10%
What investors are doing: insiders, short sellers, money flowing in or out

A price far above its fair value holds the score down, however well the company ranks. Scores read Strong from 7.0, Positive from 5.5, Mixed from 4.0 and Weak below. They describe what we measured, not what to do.

How often it updates

Ratings are worked out again after each US close and each morning, and every page says which close they are made to. US prices update every minute or two while the market is open.

Stories

Our newsroom follows official sources (company filings, statistics offices, central banks) and news outlets around the clock. When something material happens, AI writes the story from the sources it read in full, and names and links them under it. Every story is checked against our data and those sources before it is published. How we use AI.

Smart money

Insiders’, funds’ and members of Congress’s trades are shown as they were filed. Each Congress trade says how many days after the trade it was disclosed, because a trade made public a month later is old news.

What a rating is not

This is method 3.8, in use since 5 November 2026. The full method, every weight and every change since launch are in the app, under Research → Our method.

More on how we work