Distributional properties of exceedance statistics (Q1280568)

The authors consider a random variable (rv) \(X\) with distribution function (df) \(F\) and a sequence \(Y=(Y_i)_{i\geq 1}\) of independent, identically distributed rv's with common df \(G\), independent of \(X\). After setting \(\overline G=1-G\) they prove the following results, among others. Theorem 1. For any integer \(n\geq 1\) set \(S_n=\#\{j<n: Y_j\leq X\}\). Then \[ P(S_n=j)={n\choose j}E(G^j(X)\overline G^{n-j}(X)),\;j=0,1, \dots,n. \] Corollary 1. If \(F\equiv G\) then \(P(S_n=j)=1/(n+1)\). Theorem 2. For any integer \(n\geq 1\) set \(R_n=\min \{j\geq 0:S_{n+j-1}=j\), \(Y_{n+j}>X\}\), and assume that \(P(G(X)<1)>0\). Then, for \(j=0,1,\dots\), \[ P(R_n=j)={n+j-1\choose n-j} E(G^j(X)\overline G^n(X)). \] Theorem 3. Let \(K\) denote the number of records of the sequence \(Y\) not exceeding the level \(X\), and assume that \(P(G(X)<1)>0\). Then, for \(j=0,1,\dots\), \[ P(K=j)= (1/j!)E (\overline G(X)(-\log (\overline G(X)))^j I_{(0,1)}(G(X))). \] They also show that \(S_n/n\) converges in distribution to \(G(X)\) as \(n\) tends to infinity, \(R_n/n\) converges in distribution to \(G(X)/ \overline G(X)\) as \(N\) converges to infinity, providing \(G(X)<1\) a.s., and \(R_n\) has the Waring distribution \(W(n,n+1)\) if and only if \(F\equiv G\).