The Morgan Stanley gamblers ruin answer is 4/7, about 57%. Player A starts with $1, Player B starts with $2, and A wins $1 with probability 2/3 in every round. A takes all $3 a little more than half the time.
This probability question has been reported as a Morgan Stanley interview question, and it is a neat test of how you think about risk. Two players, three dollars between them, one player with a clear edge in every round. Most candidates answer 2/3 or 1/2. Both are wrong.
The Morgan Stanley Gamblers Ruin Question
Player A has $1. Player B has $2. They play rounds. In each round, A wins $1 from B with probability 2/3, and loses $1 to B with probability 1/3. The first player to hold all $3 wins the game. What is the probability that A wins?
The question is usually given with three options:
A) 2/3
B) 1/2
C) 4/7
The correct option is C. Here is the working, step by step.
Before the algebra, notice two facts. The total money never changes: it is always $3. And only A’s money matters: A can hold $0, $1, $2 or $3. At $0, B has won; at $3, A has won. The whole game lives in the two middle states.
Why the Morgan Stanley Gamblers Ruin Answer Is 4/7
Give the two unknown chances names. Let x be the probability that A eventually wins holding $1, and y the probability holding $2. The answer we want is x.
Start with x. A holds $1 and plays one round. With probability 2/3, A wins the round and moves to $2, where the chance of winning the game is y. With probability 1/3, A loses the round and drops to $0, where the chance of winning is 0. So:
x = (2/3)y + (1/3)(0) = (2/3)y
Now do the same for y. A holds $2 and plays one round. With probability 2/3, A wins the round and reaches $3, where the chance of winning the game is 1. With probability 1/3, A loses the round and drops to $1, where the chance is x. So:
y = (2/3)(1) + (1/3)x = 2/3 + (1/3)x
Two equations, two unknowns. Substitute the first into the second:
y = 2/3 + (1/3)(2/3)y = 2/3 + (2/9)y
Move the y terms together: y – (2/9)y = 2/3, so (7/9)y = 2/3, and y = (2/3)(9/7) = 6/7.
Then x = (2/3)(6/7) = 12/21 = 4/7. That is the answer: about 57%.
Check the Answer Two Ways
Good candidates check their answer instead of trusting the first pass. The classic gambler’s ruin formula gives the same result: with p the chance of winning $1 in a round, q = 1 – p, starting with i dollars and aiming for N, the chance of reaching N before 0 is (1 – (q/p)^i) / (1 – (q/p)^N). Here p = 2/3, q = 1/3, so q/p = 1/2, i = 1 and N = 3:
(1 – 1/2) / (1 – 1/8) = (1/2) / (7/8) = 4/7
Two methods, one answer. That is the sign the working is right.
The sanity checks pass too. From $2, the chance would be y = 6/7, about 86%. With fair rounds, p = 1/2, it would be 1/3, about 33%. If A won every round, it would be 1. The answer 4/7 sits between 1/3 and 1, where it should sit for a player with a 2/3 edge and a small bankroll.
| Money A starts with | Chance A takes all $3 | As a percentage |
|---|---|---|
| $0 | 0 | 0% |
| $1 | 4/7 | About 57% |
| $2 | 6/7 | About 86% |
| $3 | 1 | 100% |
The Trap Most Candidates Fall Into
The first trap is answering 2/3. That number is the chance of winning one round, not the chance of winning the game. A single lost round wipes A out, because A starts with only $1. An edge in each round does not protect A from that first-round loss.
The second trap is answering 1/2 by symmetry. That only works when the two sides are mirror images, and here they are not. A wins each round with probability 2/3, and B starts with twice as much money. B’s chance of taking everything is 3/7, about 43%, not 1/2.
The third trap is listing paths. Candidates try to write out sequences of wins and losses, but wins and losses can swing back and forth, so the list never ends. The equation method handles those loops in two lines, because the states repeat and the equations capture the repetition.
The fourth trap is forgetting the first round. A loses the whole game on round one with probability 1/3, so any answer above 2/3 is impossible before any algebra starts.
What the Interviewer Is Testing in the Morgan Stanley Gamblers Ruin Puzzle
This question tests four things, and none of them is arithmetic speed. First, it tests state thinking: naming the unknowns, writing down how the states connect, and solving. Candidates who define x and y and write the two equations are doing the job; candidates who guess are not.
Second, it checks whether you can spot a random walk: money moves up $1 or down $1, with absorbing states at $0 and $3. That shape sits behind option pricing, risk limits and trading strategies.
Third, it looks at how you verify your work. A strong candidate checks the answer with the general formula, or tests the edge cases: p = 1/2 gives 1/3, p = 1 gives 1. In a bank, an unchecked number can cost real money.
Finally, it tests whether you see the trading lesson inside the puzzle. A has an edge, winning 2 rounds in 3, and still gets ruined 3 times in 7, because the stake is too big relative to the bankroll. That is why traders size positions: a good strategy with bad sizing still ends in ruin.
Try the Morgan Stanley Gamblers Ruin With Different Numbers
Change one number and the answer moves. Each version below uses the same method.
Fair coin, A has $1 and B has $2. With p = 1/2, the formula becomes i/N, so the answer is 1/3, about 33%.
Same edge, A starts with $2 and B has $1. The answer is 6/7, about 86%: one extra dollar nearly doubles the safety, because A can now survive a lost round.
Bigger table, same edge. A has $1, B has $4, target $5. The formula gives (1 – 1/2) / (1 – 1/32) = 16/31, about 52%. B’s stack more than doubles, yet A’s chance only drops from 57% to 52%.
Without the edge, the picture flips: with a fair coin and B holding $4, A’s chance is 1/5, or 20%. Edge and bankroll size decide ruin together, never one alone.
Try Another Puzzle From the Series
If you liked the backward reasoning here, try the 253 lions puzzle: 253 perfectly rational lions, one poisoned steak, and an answer that turns on whether 253 is odd or even: Goldman Sachs 253 Lions Puzzle Answer [Solved].
For a puzzle about doubling instead of probability, try the rice chessboard puzzle: one grain on the first square, double it on every square after, and the total passes 18 quintillion grains: Rice Chessboard Puzzle Answer: 18 Quintillion Grains [Solved].
Frequently Asked Questions
Is the gamblers ruin question a real Morgan Stanley interview question?
It has been reported as a Morgan Stanley interview question, in the form used here: Player A with $1 against Player B with $2, winning $1 with probability 2/3. The firm does not publish its interview questions, so treat the attribution as reported rather than confirmed. The problem itself is a textbook classic.
What is the general gamblers ruin formula?
Start with i dollars, aim for N, and win $1 in each round with probability p (q = 1 – p). If p is not 1/2, the chance of reaching N before 0 is (1 – (q/p)^i) / (1 – (q/p)^N). If the game is fair, the chance is simply i/N: your chance equals your share of the total money.
Why is the answer not 2/3?
Because 2/3 is the chance of winning a single round, and the game needs more than one round. A loses everything on the first round with probability 1/3, so the chance of winning cannot be higher than 2/3. It is 4/7, about 57%.
How long does the game last?
On average, about 2.1 rounds. The game ends fast because A starts one loss from ruin, and most games end in one or two rounds. That is why the equation method beats listing paths.
Sources: the gambler’s ruin problem as set out in the MIT OpenCourseWare text Mathematics for Computer Science, chapter 20.1 and the Columbia lecture notes on the gambler’s ruin problem. The solution and all text are original.
Enjoying this? Add StudyFromNotes as a preferred source on Google to see our latest notes and updates more often:

Lauren Brooks is a finance professional writing on professional exams, study strategies, and finance careers.
[…] If you liked the slow-burn logic here, try the Morgan Stanley gambler’s ruin puzzle next: two players, three dollars between them, and an answer of 4/7. The method is different, equations instead of a series, but the lesson is the same: the obvious answer is usually wrong. Read it here: Morgan Stanley Gamblers Ruin Puzzle: Answer [Solved]. […]