Statistics with explicit denominators and limits

Deck and card meta now count only wins and losses in their decided totals, usage and shrinkage baseline; draws and unresolved results no longer depress that baseline. Empty samples report an unknown rate. Personal and clan Pilot Scores now share the same level inputs and qualifying population. Their responses spell out that the legacy standard-error field is an approximation, not a confidence interval for the score. The methodology documents the formulas, independent rounding of displayed rates, sample floors, dependent observations, and limits of rolling-baseline and tenure comparisons. Earlier release descriptions claiming guaranteed rankings or improvement independent of spending have been corrected.