2026-05-06: $ZTS predicted +0.75%, did -21.50%

On May 6th, the 20 picks generated a 50% directional hit rate with an average actual one-day return of -1.57%. Ten calls moved in the predicted direction, ten did not. The portfolio-level performance reflected a weak session across the sample, with positive and negative surprises distributed unevenly. The hero pick—ZTS—exemplifies this volatility. The model predicted a modest 0.75% gain with a bullish lean. The stock fell 21.5% instead, producing a 22.25 percentage point error. On the upside, IT validated the methodology's ability to identify genuine momentum: it was forecast at 1.05% and delivered 4.46%, beating the call by 3.41 points. These two outcomes sit at opposite poles of the error distribution and deserve equal weight in calibration. The week's methodology requires adjustment. A single -21.5% miss on a bullish signal—paired with a 50% directional rate—suggests either model input quality degraded yesterday, or the framework underestimates tail risk in certain securities. IT's outperformance does not offset the magnitude of the ZTS miss; it indicates the model can identify winners but struggles with downside containment in specific names. Going forward, the priority is understanding what made ZTS vulnerable to such a gap move despite a positive signal, rather than treating it as random noise. The data warrants a review of position sizing and signal confidence thresholds before the next batch.