AEGIS-QH
Estimative language

Estimative language

Words like likely and probable mean different things to different readers. We use a calibrated vocabulary, so a probability judgment means the same thing to the analyst who writes it and the executive who reads it.

What it is

A shared vocabulary for likelihood

Estimative language maps the words we use for probability to consistent bands, following the practice of allied intelligence services. It removes the guesswork from phrases like very likely, and it keeps every analyst on the same scale.

We keep two things separate: likelihood, meaning how probable something is, and confidence, meaning how much evidence stands behind the judgment.

The ladder

From almost no chance to almost certain

Almost no chance
Remote, roughly 1 to 5 percent.
Very unlikely
Highly improbable, roughly 5 to 20 percent.
Unlikely
Improbable, roughly 20 to 45 percent.
Roughly even chance
About as likely as not, 45 to 55 percent.
Likely
Probable, roughly 55 to 80 percent.
Very likely
Highly probable, roughly 80 to 95 percent.
Almost certain
Roughly 95 to 99 percent.
Why it matters

Likelihood that means the same to everyone

A calibrated vocabulary lets a board compare judgments across analysts and over time, without decoding each writer's private sense of probable.

We never use numbers to imply a precision we do not have. The bands are a guide to meaning, not a false claim of exactness.

See our full standards
  • A fixed ladder of probability words
  • Likelihood kept separate from confidence
  • No false numeric precision
  • Consistent across analysts and time
Method and technology

Explore the standards and technology behind every finding.

Source reliability Information credibility Estimative language Provenance The human gates Artificial Intelligence Quantium Computing
Engage

Due diligence welcome

Vetted counterparties can review our standards, governance, and methodology in depth under appropriate confidentiality.

Request the detail Trust & compliance