Bagel Defects Expected Count

This quant interview question is framed as a simple production-line story but is really about stochastic dependence over time. Instead of independent trials, the quality of each item depends on the previous one, giving the problem a Markov chain flavor. In quant prep, this kind of setup is common because many market models have state-dependent dynamics rather than identically distributed steps.

It trains your understanding of expected values in a dependent sequence, together with comfort formalizing a process into states and transition probabilities. You need to reason about waiting times, conditional expectations, and how to express an unknown expectation in terms of itself via a recursive relationship. That blend of probability intuition and algebraic discipline is central to strong quant interviews.

This matters because trading strategies, execution algorithms, and risk systems often evolve in a path-dependent way. Interviewers want to see you quickly convert a verbal description into a probabilistic model, identify the relevant expectations, and manipulate them cleanly under time pressure. Problems like this reveal whether your quant interview preparation has gone beyond plug-and-chug formulas into genuine stochastic thinking.

What it tests

Problems involving waiting times or counts until a certain event in Markov chains are governed by the law of total expectation, which leverages the chain's memoryless property at each state. The expected time to reach a target state can be decomposed into cases based on the outcome of the next step, with each case weighted by its transition probability. This recursive structure arises because, after each transition, the process 'restarts' from the new state, so the expected remaining time is the same as if you had started there. The principle holds because Markov chains have no memory beyond the current state, making the future evolution depend only on where you are now, not on how you got there. This allows you to write equations for expected times that reference themselves, leading to solvable systems.