# Stop Rule Inequalities for Uniformly Bounded Sequences of Random Variables

Theodore P. Hill and Robert P. Kertz
Transactions of the American Mathematical Society
Vol. 278, No. 1 (Jul., 1983), pp. 197-207
DOI: 10.2307/1999311
Stable URL: http://www.jstor.org/stable/1999311
Page Count: 11

You are not currently logged in.

Access your personal account or get JSTOR access through your library or other institution:

## Abstract

If $X_0, X_1,\ldots$ is an arbitrarily-dependent sequence of random variables taking values in [0,1] and if $V(X_0,X_1,\ldots)$ is the supremum, over stop rules $t$, of $EX_t$, then the set of ordered pairs $\{(x, y): x = V(X_0, X_1,\ldots, X_n)$ and $y = E(\max_{j\leqslant n}X_j)$ for some $X_0,\ldots, X_n\}$ is precisely the set $$C_n = \{(x, y): x \leqslant y \leqslant x\big(1 + n(1 - x^{1/n})\big); 0 \leqslant x \leqslant 1\};$$ and the set of ordered pairs $\{(x, y): x = V(X_0,X_1,\ldots)$ and $y = E(\sup_n X_n)$ for some $X_0, X_1,\ldots\}$ is precisely the set $$C = \bigcup_{n = 1}^\infty C_n$$. As a special case, if $X_0, X_1,\ldots$ is a martingale with $EX_0 = x$, then $E(\max_{j\leqslant n} X) \leqslant x + nx(1 - x^{1/n})$ and $E(\sup_n X_n) \leqslant x - x\ln x$, and both inequalities are sharp.

• 197
• 198
• 199
• 200
• 201
• 202
• 203
• 204
• 205
• 206
• 207