Extremal characteristics of conditional models

Abstract Conditionally specified models are often used to describe complex multivariate data. Such assume implicit structures on the extremes. So far, no methodology exists for calculating extremal characteristics of conditional since copula and marginals not expressed in closed forms. We consider bivariate that specify distribution X Y . provide tools quantify assumptions extremes this class models. In particular, these allow us approximate tail coefficient asymptotic independence $$\eta$$ η apply methods a widely model wave height period. Moreover, we introduce new condition parameter space Heffernan Tawn (Journal Royal Statistical Society: Series B (Methodology) 66(3), 497-547, 2004), prove does capture , when $$\eta <1$$ < 1