On large deviations of self-normalized sum (Q2769658)

Let \(X_1,X_2,\ldots\) be i.i.d. random variables with \(\text{ E}X=0\). The so-called self-normalized sum \(t_n\) is defined by \(t_n= (X_1+\cdots+X_n)/(X^2_1+\cdots+X^2_n)\). The large deviations theorem is proved for the self-normalized sum. The proof is based on the cumulant method of \textit{V. P. Leonov} and \textit{A. N. Shiryaev} [Theory Probab. Appl. 4, 319-329 (1960); translation from Teor. Veroyatn. Primen. 4, 342-355 (1959; Zbl 0087.33701)] and \textit{L. Saulis} and \textit{V. Statulevičius} [``Limit theorems of large deviations'' (1991; Zbl 0744.60028)].NEWLINENEWLINEFor the entire collection see [Zbl 0968.00043].