Resolution
Resolution is the ability of forecasts to meaningfully separate situations where an event is more likely from situations where it is less likely. Higher resolution means your probabilities vary appropriately across cases.
Definition
Resolution captures how much your forecasts differentiate between different situations. A forecaster with high resolution assigns higher probabilities in cases where events occur and lower probabilities where they do not.
Intuition
If you always predict 50%, you have zero resolution, even if you might look “okay” on average in some datasets. Resolution reflects whether your probabilities move with information.
Relation to calibration
Resolution is not the same as calibration. You can be calibrated but have low resolution (always near base rates), or have high resolution but be miscalibrated if you push probabilities too far.
Why it matters
Better resolution tends to improve Brier score and Brier skill score because it helps you capture real differences between cases rather than averaging everything together.
Related
Resolution is often discussed alongside reliability and Brier decomposition, and it relates closely to sharpness.