How each of the 14 scoring factors has historically predicted outcomes. Signals are ranked by their predictive lift — the gap between their average sub-score on winning picks vs. losing picks.
Predictive Lift — the difference between a signal's average sub-score on winning picks vs. losing picks. A positive lift means the signal was higher on days the pick went up, which is a sign it's genuinely informative. A lift near zero means the signal didn't differentiate wins from losses.
Strong vs Weak — picks are split at the median signal score. "Strong" means the signal was in the upper half on that pick's day; "Weak" means the lower half. A useful signal should have a noticeably higher win rate and average return in the Strong bucket.
Trend — compares win rate for the most recent 10 picks vs the prior 10. Shows whether a signal is becoming more or less reliable over time.
Note: these statistics are based on live pick history, not the backtest. Small sample sizes will produce noisy estimates — treat any signal with fewer than 20 picks with extra caution.