4
$\begingroup$

Consider a positive random variable $X$ and call $E[X]$ its expectation. For any positive $a \in \mathbb{R}$, an upper bound for the probability of $P(X>a)$ is provided by the Markov's Inequality, $$ P(X>a) \leq \frac{E[X]}{a}, $$ Is there an analogous lower bound that is based only on the knowledge of the expectation?

$\endgroup$
3
$\begingroup$

Consider a distribution with $$P(X=m\delta)=1-\delta$$ $$P\left(X=m\tfrac{1-\delta+\delta^2}{\delta}\right)=\delta$$ then $E[X]=m$ but $P(X\gt a )$ can be made arbitrarily small by choosing $\delta \lt \frac{a}{m}$ and if necessarily smaller.

So to get something useful, you would need some constraint on $X$ such as a given maximum or a given variance.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.