Bayesian Posterior That k Red Parts Came From Factory A

Read the full interview experience this question came from →

Quick Overview

A probability question on Bayesian updating: one of two factories with different red-part rates is chosen at random, and k parts drawn from it are all red. It asks for the conditional probability that the parts came from the first factory, testing likelihoods, priors and careful evaluation of powers.

Bayesian Posterior That k Red Parts Came From Factory A

Company: Sig

Role: Quantitative Trader

Category: Statistics & Math

Difficulty: medium

Interview Round: Online Assessment

Two factories make the same part in two colors. Factory A's output is $a\%$ red and $(100-a)\%$ black. Factory B's output is $b\%$ red and $(100-b)\%$ black. You pick one of the two factories at random, each with equal probability, and draw $k$ parts from its output. All $k$ parts turn out to be red. What is the conditional probability that these $k$ parts came from factory A? Give a general expression in $a$, $b$ and $k$, then evaluate it for the values you are given. The assessment generates the specific numbers randomly, and each question states whether the answer is an integer or a decimal rounded to two places. ```hint Two competing explanations Work out how plausible an all-red sample of this size is under each factory before you bring in how the factory was chosen. ``` ### Constraints and Clarifications - $a$ and $b$ are percentages between 0 and 100, and $k$ is a positive integer. - Each factory is chosen with probability one half, and all $k$ parts come from the chosen factory. - Treat each factory's output as very large, so the colors of the drawn parts are independent given the factory. ### Clarifying Questions - Is each factory equally likely to be chosen, or does one of them supply more of the parts? - Are the parts drawn from a large production run, or from a small batch of known size where drawing without replacement matters? - Should rounding happen only at the end, and to how many decimal places? ### What a Strong Answer Covers - Applying Bayes' theorem with the equal prior, rather than reporting the red rates themselves - The probability of an all-red sample of size $k$ under each factory - A simplified closed form, and a way to evaluate it without overflow or underflow when $k$ is large - Degenerate inputs, such as equal red rates or a factory that makes no red parts ### Follow-up Questions - How does the posterior change if exactly $j$ of the $k$ parts are red? - If factory A supplies a larger share of the parts than factory B, how does the answer change? - When $a$ is greater than $b$, how many consecutive red parts are needed before the posterior for factory A exceeds a target level $c$?

Overview: A probability question on Bayesian updating: one of two factories with different red-part rates is chosen at random, and k parts drawn from it are all red. It asks for the conditional probability that the parts came from the first factory, testing likelihoods, priors and careful evaluation of powers.

Read the full Sig Quantitative Trader interview experience this question came from

|Home/Statistics & Math/Sig
Sig logo
Sig
Sep 19, 2026
mediumQuantitative TraderOnline AssessmentStatistics & Math
0
0

Two factories make the same part in two colors. Factory A's output is a%a\% red and (100−a)%(100-a)\% black. Factory B's output is b%b\% red and (100−b)%(100-b)\% black. You pick one of the two factories at random, each with equal probability, and draw kk parts from its output. All kk parts turn out to be red.

What is the conditional probability that these kk parts came from factory A? Give a general expression in aa, bb and kk, then evaluate it for the values you are given. The assessment generates the specific numbers randomly, and each question states whether the answer is an integer or a decimal rounded to two places.

Constraints and Clarifications

  • aa and bb are percentages between 0 and 100, and kk is a positive integer.
  • Each factory is chosen with probability one half, and all kk parts come from the chosen factory.
  • Treat each factory's output as very large, so the colors of the drawn parts are independent given the factory.

Clarifying Questions Guidance

  • Is each factory equally likely to be chosen, or does one of them supply more of the parts?
  • Are the parts drawn from a large production run, or from a small batch of known size where drawing without replacement matters?
  • Should rounding happen only at the end, and to how many decimal places?

What a Strong Answer Covers Guidance

  • Applying Bayes' theorem with the equal prior, rather than reporting the red rates themselves
  • The probability of an all-red sample of size kk under each factory
  • A simplified closed form, and a way to evaluate it without overflow or underflow when kk is large
  • Degenerate inputs, such as equal red rates or a factory that makes no red parts

Follow-up Questions Guidance

  • How does the posterior change if exactly jj of the kk parts are red?
  • If factory A supplies a larger share of the parts than factory B, how does the answer change?
  • When aa is greater than bb , how many consecutive red parts are needed before the posterior for factory A exceeds a target level cc ?
Loading comments...