The source composite averages five assigned ratings with equal weight. Hypothetical all ninety ratings produce ninety; changing one rating to one hundred produces ninety-two. Bars share four pixels per composite point. This arithmetic example does not establish that one hundred is selectable or that the ratings measure real athlete ability. Because five values are averaged, raising one by ten raises the composite by two; raising one by five raises it by one. Bars share one hundred pixels per composite-point change. There are no independently adjustable skill weights in this formula despite broader page language about weighting. The displayed dominance percentage is composite divided by ninety-nine times one hundred, to one decimal. Hypothetical composites ninety and ninety-nine produce ninety point nine and one hundred percent. Bars share three pixels per displayed percentage point. This is a linear normalization, not an exact radar polygon area percentage. Full native JSON/reset, responsive and first-footer checks preserve source ratings and fields; clipboard summary and athlete facts are independently unverified.