Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top
  1. If $X$ and $Y$ are dependent random variables, then it is possible that $Var(X+Y) > Var(X) + Var(Y)$.

I only know that the two are equal for independent random variables; for dependent variables, would this be the case?

  1. According to forecasts, the end-of-year value in dollars of IBM stock has variance 10. If an investor holds a portfolio containing 5 shares of IBM stock and 240 dollars of idle cash, what is the variance in the end-of-year value in dollars of his portfolio?

I would assume it's 50, as variance is additive?

  1. If X and Y are random variables such that $P(X=0)=0.5$ and $P(Y=0)=0.1$, then is $P((X+Y)/2=0)$ equal to $P(X=0)/2 + P(Y=0)/2 = 0.3$?

I don't believe that it's possible to add probabilities like this, would I multiply instead?

share|cite|improve this question
In general, $\operatorname{var}(X+Y) = \operatorname{var} X + \operatorname{var} Y + \operatorname{Cov}(X,Y)$, where the last term is the covariance. If $X$ and $Y$ are positively correlated ("$X$ and $Y$ increase or decrease together"), then this term will be positive. – Clement C. Mar 5 at 23:42
@Clement You are missing a factor of $2$ in front of $cov(X,Y)$ or $cov(X,Y)+(cov(X,Y))^T$ in the multi-dimensional case. – A.S. Mar 5 at 23:59
Oh, true -- thanks for pointing out. And I was going to write something along the line of "it's easy to remember, as it mimics the formula for $(a+b)^2=a^2+b^2+2ab$ with the cross-term"... yes, easy to remember indeed. – Clement C. Mar 6 at 0:05
up vote 1 down vote accepted

1: $\text{Var}(Z) = \text{E}\Big[\big(Z - \text{E}(Z)\big)^2\Big]$. It follows that $$\begin{align} \text{Var}(X+Y) &= \text{E}\bigg[\big(X+Y - \text{E}(X) - \text{E}(Y)\big)^2\bigg] \\ &= \text{E}\bigg[\big(X - \text{E}(X)\big)^2 + \big(Y - \text{E}(Y)\big)^2 + 2\big(X - \text{E}(Y)\big)\big(Y-\text{E}(Y)\big)\bigg] \\ &= \text{Var}(X) + \text{Var}(Y) + 2\text{Cov}(X,Y). \end{align}$$ So if the covariance is positive, the variance of $X+Y$ exceeds $\text{Var}(X) + \text{Var}(Y)$.

2: No, it's 250. You have that $\text{Var}(aX) = a^2\text{Var}(X)$, and therefore the variance of $5$ stocks with variance $10$ is $5^2\cdot 10 = 250$. Technically, you then need to consider the variance of those 240 dollars. But since their value doesn't have any uncertainty, that variance is zero.

3: You're correct that the answer is "no". Its not even possible to compute the probability precisely, because $X+Y$ can be zero even if $X$ and $Y$ are both non-zero. So $P(X=0)\cdot P(Y=0)$ is only a lower bound for the probability $P(X+Y=0)$.

Edited: The answer previously stated that 50 is the correct answer for question 2, which was wrong.

share|cite|improve this answer
@Teepeemm Uh, yes, of course. Fixed. – fgp Mar 6 at 8:40

Suppose that $Y=X$. Then the variance of $X+Y$ is the variance of $2X$, which is $4$ times the variance of $X$.

So for example if $X=1$ if when tossing a fair coin we get a head, and $X=0$ otherwise, and $Y=X$, then $\text{Var}(X+Y)\gt \text{Var}(X)+\text{Var}(Y)$.

share|cite|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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