Mean Dice Roll Surprises

This quant interview question is about understanding the distribution of the minimum of several discrete random variables and linking it to expected value. It sits at the crossroads of probability and combinatorics, and is very representative of the kind of discrete expected value puzzles that appear in quant interviews and serious quant prep resources. The setting is simple on the surface, but the structure of the random outcome space is rich enough to test genuine insight.

It trains your grasp of distributions of order statistics, conditional probability, and careful event counting in a finite probability space. You need to recognize how to characterize events about the minimum in a clean, global way, and then translate that into the expected value via its distribution. It also reinforces comfort manipulating powers, complements, and differences of probabilities.

This matters for quant interviews because understanding minima and maxima under independence is central to risk, extremes, and tail events. Interviewers at top quant firms want to see that you can move fluidly from an informal description to a rigorous probability structure. Problems like this reveal whether your quant prep has built genuine probabilistic thinking, not just formula memorization.

What it tests

When dealing with the distribution of the minimum (or maximum) of several independent, identically distributed discrete random variables, the key is to count the number of outcomes where all variables are at least a certain value, and then subtract those where all are strictly greater. This exploits the structure of the outcome space: for $n$ dice each with $d$ sides, the set of outcomes where all dice are at least $x$ forms a sub-cube of size $(d - x + 1)^n$. The probability that the minimum is exactly $x$ is the difference in the sizes of these sub-cubes for $x$ and $x+1$. This approach generalizes to any such minimum/maximum problem because the minimum is determined by the lowest threshold all variables simultaneously meet, and the difference isolates the cases where $x$ is the lowest value present. The principle holds because the event 'minimum is at least $x$' is monotonic and easy to count, while 'minimum is exactly $x$' is the difference between two such monotonic events.