2026-05-21: $HPQ predicted -2.06%, did +15.25%

On May 21, 2026, the pattern-based model generated 20 picks with a directional accuracy of 50% (10 correct). The portfolio averaged a gain of 1.17% across all selections. The range was wide: best single pick returned 15.25%, worst returned −3.02%. HPQ was flagged as bearish with a predicted decline of 2.06%, but instead rose 15.25%—a 17.31 percentage point miss in the opposite direction. This reversal was the largest absolute error of the day and became the "hero" pick by virtue of its outsized actual return, not its predictive merit. The model was wrong on direction, yet the position would have been profitable at scale. JD was predicted up 1.04% but fell 3.02%, confirming that misses occurred across both directional calls. The 50% directional hit rate and wide error dispersion suggest the methodology is not reliably capturing intraday momentum or mean-reversion signals in this sample. HPQ's sharp reversal against a bearish signal indicates the model may be sensitive to gap-up events or overnight news that overwhelms technical pattern logic. With half the picks failing to predict direction and occasional large reversals, the calibration story this week is one of volatility outpacing signal strength. The presence of a profitable-but-misdirected pick serves as a reminder that positive returns and accurate predictions are not synonymous.