2026-05-18: $GLNG predicted +1.06%, did -1.99%

On May 18th, the 20-pick set delivered directional accuracy of 7 out of 20 (35%), with a portfolio average move of −0.16%. The range spanned from HALO's +1.91% gain to GLNG's −1.99% loss. Across the full set, picks were distributed between bullish and bearish calls, but the net result underperformed a neutral baseline. The hero pick was GLNG, flagged as a bullish setup with a predicted one-day move of +1.06%. Instead, GLNG closed down 1.99%, creating an absolute error of 3.05 percentage points—the largest miss of the day. This was also the worst individual pick in the set. The model's directional confidence on this call proved misplaced; what was positioned as an upside candidate moved sharply in the opposite direction. The 35% directional hit rate and the magnitude of the GLNG miss suggest the methodology encountered a calibration gap yesterday. Seven correct calls out of twenty is below a 50/50 coin-flip benchmark, indicating the pattern signals generated did not translate to reliable prediction. The hero pick's failure—a 3+ point error on a 1.06% bullish thesis—is the most concrete evidence that yesterday's setup assumptions broke down. No extrapolation about today's reliability is warranted; this is simply what the data recorded and what needs to be incorporated into how the model assesses similar patterns this week.