Average Rolls to See All Die Faces

This classic expected value puzzle is about how long it takes, on average, for randomness to generate a complete set of distinct outcomes. It sits at the intersection of probability theory and quant prep, and is a canonical example of the coupon collector framework. Candidates see variants of this in many quant interviews when firms want to probe intuition about random processes and waiting times.

It trains your understanding of discrete distributions, linearity of expectation, and how expectations behave under changing probabilities. You practise structuring a stochastic process into stages, quantifying each stage, and combining them into a single expectation. It also reinforces comfort with infinite sums and approximations that are standard tools in quantitative finance.

This matters in quant interviews because it mirrors real quant work: reasoning about rare events, convergence, and sampling efficiency. It tests whether you can translate an informal story into a precise probabilistic model, which is central in quantitative research, systematic trading, and risk modeling. For quant prep, mastering this question class builds a strong foundation for more complex interview problems involving stochastic processes and Monte Carlo methods.

What it tests

This problem class is governed by the coupon collector principle, which models the process of collecting all distinct items from a set when each trial yields a random item. The key structure is that the probability of obtaining a new, unseen item decreases as more unique items are collected, making each subsequent 'collection' take longer on average. The expected time to complete the collection is the sum of expected times to collect each new item, where each stage is a geometric random variable with success probability proportional to the number of remaining unseen items. This additive structure arises from the independence of each trial and the memoryless property of geometric distributions. The pattern holds because, at each stage, the process resets with a smaller pool of 'needed' outcomes, and the expected waiting time for each new outcome is inversely proportional to how many remain.