Positive returns and upcrossings in a random walk (Q5896251)

Consider samples of size m(n) from a continuous distribution F(G); now pool the two samples and arrange the resulting \(n+m\) observations in ascending order; call \(\beta_ i\) equal to \(+1\) (-1) if the i-th observation comes from F(G). A random walk is then obtained by putting \(S_ 0=0\), while for positive i, \(S_ i=\beta_ 1+...+\beta_ i.\) The author investigates a number of rank order statistics such as the number of positive returns, the number of upcrossings, the number of runs of positive returns, etc. The paper is closely related to \textit{M. Dwass}, Ann. Math. Stat. 38, 1042-1053 (1967; Zbl 0162.502) where the case \(m=n\) has been treated.