What is confidence?
Confidence is a measure of how reliable the data behind a signal is. A signal drawn from a large sample of recent activity is more confident than one drawn from a small sample, even if both point the same way. It is the engine’s read on how much to trust the signal.
Why does confidence matter?
Because a pattern from thin data can be noise. Confidence lets the engine deprioritise signals that might be real but are not yet well-established, so low-confidence candidates step aside when stronger, better-supported ones are available. It is a guard against acting on too little.
How is confidence different from proof strength?
Confidence is about reliability — how trustworthy the sample is; proof strength is about magnitude — how much evidence there is. A product can have strong raw numbers from an unreliable window, or modest numbers from a solid one. The message score weighs both.
How does confidence feed the message score?
It is one of the score’s components. When two candidates are otherwise comparable, the more confident one scores higher and wins the slot, so the messages that surface are the ones the data most firmly supports.
See the engine in action
The engine weighs how reliable each signal's data is, so well-supported messages win.