ISS 2017 Statistics Paper-1 Solution: Question 12 (Drunkard's Walk Probability)
Continuing our ongoing question-by-question walkthrough of ISS Statistics Paper-1 previous year papers, we now move to the 2017 paper and tackle Question 12 — a lovely little problem on random walks that trips up a lot of first-time aspirants. As always, we will build the solution from the ground up rather than jumping to the formula.
Quick Summary
Topic: Discrete probability — binomial distribution applied to a 1-D random walk (drunkard's walk)
Question Reference: ISS 2017, Statistics Paper-1, Question 12
Correct Answer: Option (d) — 462(pq)5
The Question
A drunkard takes a forward step with probability p and a backward step with probability q. After taking 11 steps, the probability that he is one step away from the starting point is:
(a) p6 + q6
(b) 2(p6 + q5)
(c) 462 p6q5
(d) 462(pq)5
Step 1: Set Up the Random Walk
Picture the drunkard standing at position 0 on a number line. Every step he takes is independent of the others, and each step is either:
A forward step (+1), which happens with probability p, or
A backward step (−1), which happens with probability q, where p + q = 1 (since these are the only two possibilities for any single step).
He takes a total of n = 11 such steps. We want the probability that after all 11 steps, his net displacement from the starting point is exactly 1 step away — that could mean he ended up at position +1 or at position −1. Both of these count as being one step away from the starting point, so we must add both probabilities together.
Step 2: Express Net Displacement in Terms of Number of Forward Steps
Let k be the number of forward steps out of the 11 total steps. Then the number of backward steps is automatically (11 − k), because every step is either forward or backward.
His net displacement from the start is simply:
Net displacement = (number of forward steps) − (number of backward steps) = k − (11 − k) = 2k − 11
We want this net displacement to equal +1 or −1 (both mean one step away). So we solve two small equations:
2k − 11 = 1 ⟹ 2k = 12 ⟹ k = 6
2k − 11 = −1 ⟹ 2k = 10 ⟹ k = 5
So the drunkard ends up one step away from the start in exactly two scenarios: either he took 6 forward steps and 5 backward steps (ending at +1), or he took 5 forward steps and 6 backward steps (ending at −1).
Step 3: Model This as a Binomial Distribution
Since each of the 11 steps is an independent trial with only two outcomes (forward with probability p, backward with probability q), the number of forward steps k out of 11 trials follows a Binomial distribution with parameters n = 11 and success probability p.
The binomial probability formula for getting exactly k successes out of n trials is:
P(X = k) = C(n, k) · pk · q(n−k)
where C(n, k) = n! / [k!(n−k)!] is the number of ways to choose which k of the n steps are forward steps.
Step 4: Compute the Probability for k = 6 (Ending at +1)
Here n = 11 and k = 6, so the number of backward steps is 11 − 6 = 5.
P(k = 6) = C(11, 6) · p6 · q5
Let's compute C(11, 6) by hand:
C(11, 6) = 11! / (6! · 5!) = (11 × 10 × 9 × 8 × 7) / (5 × 4 × 3 × 2 × 1)
Working out the numerator: 11 × 10 = 110, 110 × 9 = 990, 990 × 8 = 7920, 7920 × 7 = 55440.
The denominator is 5! = 120.
So C(11, 6) = 55440 / 120 = 462.
Therefore: P(k = 6) = 462 · p6 · q5
Step 5: Compute the Probability for k = 5 (Ending at −1)
Here n = 11 and k = 5, so the number of backward steps is 11 − 5 = 6.
P(k = 5) = C(11, 5) · p5 · q6
A useful shortcut from combinatorics — the symmetry property of binomial coefficients — tells us that C(n, k) = C(n, n−k). Here that means C(11, 5) = C(11, 6) = 462. You can verify this directly too:
C(11, 5) = 11! / (5! · 6!), which is exactly the same fraction as C(11, 6) = 11! / (6! · 5!) — multiplication in the denominator doesn't care about order, so both equal 462.
Therefore: P(k = 5) = 462 · p5 · q6
Step 6: Add the Two Probabilities Together
Since ending at +1 and ending at −1 are mutually exclusive events (he cannot be at both positions at once), we add their probabilities to get the probability of being one step away from start:
P(one step away) = P(k = 6) + P(k = 5) = 462 p6q5 + 462 p5q6
Now factor out the common terms. Both expressions share a factor of 462, p5, and q5:
= 462 · p5 · q5 · (p + q)
Pro Tip: This is the single trickiest step in the whole problem, and it's exactly where most students lose marks under exam pressure — forgetting that p + q = 1 and leaving the answer in an unsimplified two-term form that doesn't match any option. Whenever an exam problem gives you two complementary probabilities that must add to 1 (like forward and backward, or success and failure), always check whether substituting p + q = 1 collapses your expression into something neater. The habit of automatically scanning for this kind of simplification — rather than stopping once you've technically got an answer — is exactly the kind of drilled reflex that systematic MCQ practice, the sort built into structured coaching for exams like ISS, is designed to make automatic.
Since p + q = 1, this simplifies beautifully:
= 462 · p5 · q5 · 1 = 462 · (pq)5
Step 7: Final Answer
So the probability that the drunkard is one step away from the starting point after 11 steps is:
462(pq)5
This matches option (d)exactly. Checking against the official ISS 2017 Statistics Paper-1 answer key (Series A), item 12 is listed as D — so our independently derived answer is fully confirmed by the official key.
Why This Question Matters
This question is a textbook test of whether you can translate a word problem into a binomial random variable correctly. The real skill being tested isn't memorizing the binomial formula — it's recognizing that being one step away is not a single outcome but the union of two symmetric outcomes (+1 and −1), and then having the algebraic discipline to simplify using p + q = 1 rather than leaving a messy two-term expression. Random walk problems of this flavor appear repeatedly across ISS, IIT JAM (MS), and GATE (ST) papers, almost always dressed up in different real-world stories (gambler's ruin, particle movement, stock price ticks) while testing the exact same binomial-counting idea underneath.
Frequently Asked Questions
Why do we add the probabilities for k = 6 and k = 5 instead of just picking one?
Because being one step away from the starting point includes both being at position +1 and position −1 — these are two separate, non-overlapping ways to satisfy the condition. Whenever a probability question describes a condition that multiple distinct outcomes can satisfy, you must add up the probabilities of all of them, not just one.
Why is the number of forward steps binomially distributed here?
Because each of the 11 steps is an independent trial with exactly two possible outcomes (forward or backward), and the probability of a forward step (p) stays the same on every trial. This is precisely the definition of a binomial setup: a fixed number of independent, identically distributed Bernoulli trials.
Is there a quicker way to see that C(11, 6) equals C(11, 5)?
Yes — this is the symmetry property of binomial coefficients, C(n, k) = C(n, n − k). Intuitively, choosing which 6 of the 11 steps are forward steps is equivalent to choosing which 5 of the 11 steps are backward steps, so both counts must be identical.
What would change if the drunkard took an even number of steps instead of 11?
With an even number of steps, say 10, the net displacement 2k − 10 can never equal an odd number like +1 or −1 for integer k — it can only be even. So being one step away would actually be an impossible event with probability zero after an even number of steps. This parity argument is itself a common trick used in harder variants of this question.
Why can't the answer be option (c), 462 p6q5, since that's part of our working?
Option (c) only captures the probability of ending at +1 (the k = 6 case) and ignores the symmetric possibility of ending at −1 (the k = 5 case). Since the question asks for one step away without specifying direction, both cases must be included, which is exactly why the final simplified answer turns out to be 462(pq)5 rather than the single unsimplified term in option (c).
Does this same logic apply to the classic Gambler's Ruin problem?
Yes, the underlying mechanics are identical — a gambler's position after n independent bets, each a win or loss, is governed by the same binomial net-displacement logic used here. The difference is that Gambler's Ruin problems typically ask about absorbing boundaries (going bankrupt or reaching a target), which adds a layer of recursive probability on top of this basic counting idea.
If anything in this derivation felt unclear, or you spotted a quicker route to the same answer, drop a comment below — and if this helped clear things up, consider sharing it with a fellow ISS aspirant who's working through the same 2017 paper.

Comments