Prediction Accuracy

Directional accuracy compares the price move with the saved verdict at the evaluation horizon: undervalued is correct when price rises, overvalued when it falls, and fair when the absolute move is under 5%.

MAPE (Mean Absolute Percentage Error) measures the gap between the saved fair-value estimate and the observed price at the evaluation horizon, relative to the observed price.

Fair-value hit rate checks the observed price at the evaluation horizon: undervalued counts as a hit at or above fair value, overvalued at or below, and fair within 5%.

For each saved prediction, the evaluator uses the first recorded price on or after its 30-, 90-, or 365-day target date. The scorecard reports results only for outcomes that have reached their evaluation date.

These are historical prediction comparisons, not simulated portfolio returns or actual trading performance. A zero sample count means there is no outcome data to evaluate yet. Past performance does not guarantee future results.