Two Envelopes Paradox Answer: Why Switching Gains Nothing [Solved]

It does not matter whether you switch. Keep your envelope. The two envelopes paradox answer is that switching gains nothing: the famous argument that the other envelope is worth 1.25 times yours hides one small error.

Here is the question, as reported from a Graduate Trader interview at Vivienne Court and as it appears in standard quant interview collections. Two identical envelopes sit on a table. One holds twice as much money as the other. You pick one at random, and before you open it, you are offered the chance to switch. Stay, switch, or does it not matter?

The Two Envelopes Paradox Answer in Plain Words

Call the smaller amount A and the larger 2A. You picked at random, so you hold A half the time and 2A half the time. The expected value of your envelope is half of A plus half of 2A, which is 1.5A. The other envelope works the same way. When you hold A, it holds 2A. When you hold 2A, it holds A. Its expected value is also 1.5A. Both envelopes are worth exactly the same, so switching changes nothing.

Where the 1.25x Argument Goes Wrong

Most candidates answer “switch”, and show working. Call the amount in your envelope x, they say. The other envelope holds either 2x or x/2, each with probability one half. Its expected value is therefore 0.5 times 2x plus 0.5 times x/2, which is 1.25x. Switching looks like a 25 percent gain on every deal, so always switch.

The argument is smooth, and it is wrong in exactly one place. The letter x is doing two jobs. In “2x”, x stands for the smaller amount. In “x/2”, x stands for the larger amount. Those are two different numbers, yet the calculation adds them as if they were one fixed number. Write the two cases out in full and the error is plain to see.

What you hold	The other envelope holds	Your gain from switching
Smaller amount, A	2A	Gain A
Larger amount, 2A	A	Lose A
Average gain		Zero

Switching gains A in one case and loses A in the other. The two cases are equally likely, so the average gain is zero. The 1.25x calculation never writes down what x is in each case: it treats x as a fixed amount, then lets it quietly change value halfway through the sum.

The Two Envelopes Paradox Answer, Worked With $100 and $200

Numbers make the error concrete. Suppose the smaller amount is $100 and the larger is $200. You pick at random. Half the time you hold $100, and the other envelope holds $200, so switching gains you $100. Half the time you hold $200, and the other holds $100, so switching loses you $100. Your average gain is half of $100 minus half of $100, which is zero. Each envelope is worth $150 on average. You cannot beat a coin flip by swapping its sides.

Now run the naive argument with the same numbers and watch it break. It says the other envelope is worth 0.5 times 2x plus 0.5 times x/2. But no single x is both $100 and $200. If x is $100, you hold the smaller amount, and the other envelope holds exactly $200, not “either $200 or $50”. If x is $200, you hold the larger amount, and the other holds exactly $100, not “either $400 or $100”. The formula plugs one fixed x into both branches, but the branches describe two different candidates. In the real puzzle, once you hold an envelope, the other one is fixed. The 50/50 chance is over which envelope you picked, not over what the other one holds.

The Switch-Back Test That Kills the Argument

There is a faster way to see the flaw, and it is the line to use first in an interview. If the 1.25x argument were valid, it would apply to whichever envelope you hold. Switch to the second envelope, run the same argument, and it tells you to switch back to the first. Run it again and it tells you to switch again, forever. An argument that proves each envelope is better than the other proves nothing at all. Say this before you do any arithmetic.

Compare this with the Monty Hall problem, where switching really does win. There the host knows where the car is and always opens a door with a goat behind it, and that knowledge breaks the symmetry between your door and the other one. In the envelopes puzzle nothing breaks the symmetry. No one knows anything, no new information arrives, and no choice you make can turn a fair coin flip into free money. See the Monty Hall problem answer for the contrast: the puzzle where switching is right, and the reason it works there is exactly the reason it fails here.

When Switching Actually Is the Right Move

The two envelopes paradox answer stays the same for the question as stated, but it rests on one condition: you know nothing about how the amounts were chosen. Suppose you open your envelope and find $200, and you know the smaller amount is always between $10 and $100. Then $200 cannot be the smaller amount, so you hold the larger one, and the other envelope holds $100. Keep it. Suppose instead you open it and find $50. Then you could hold the smaller amount, with $100 in the other envelope, or the larger amount, with $25 in the other. With the smaller amount spread evenly between $10 and $100, those two cases are equally likely, so the other envelope is worth $62.50 on average against your $50. Switch.

That is the honest resolution. It is not a trick of arithmetic. It is a question about missing information. The paradox version gives you no distribution, so there is nothing to condition on, and switching gains nothing. The moment you know the range, or you see the amount, the symmetry breaks and the correct move depends on what you see. A strong candidate says this out loud: “As stated, it does not matter. If I knew the distribution, I would compute the conditional expected value.”

What the Interviewer Is Testing in the Two Envelopes Paradox

The two envelopes paradox answer rewards five skills at once.

1. Whether you test a slick argument or accept it. The 1.25x calculation looks like real mathematics. Your interviewer wants to see whether you check what each symbol means before you trust the result. The candidate who asks “what is x, exactly?” beats the candidate who copies the formula.

2. Careful expected value. Expected value is only as good as the probabilities behind it. Here the random event is which envelope you picked, and the calculation must be built on that event. Candidates who attach probabilities to the wrong thing get a confident wrong answer.

3. Symmetry reasoning. The two envelopes are identical in every way that matters, so no argument can make one of them better than the other. Use symmetry to rule out answers before you compute.

4. Comfort with missing information. The question gives you no distribution over the amounts. A good candidate names that gap instead of inventing one.

5. Communication under pressure. State the answer first, then the reason. “It does not matter, and here is why” is a stronger opening than two minutes of algebra that ends in a shrug.

Try Another Puzzle Next

If this one made you pause, try the Monty Hall problem answer next: three doors, one car, and a switch that really does double your chances. Then test your probability instincts against the birthday paradox answer, where just 23 people in a room make a shared birthday more likely than not.

Is the two envelopes paradox answer really to stay?

Yes. Staying and switching have the same expected value, so neither move beats the other. The two envelopes paradox answer is that the 1.25x argument fails because it uses one letter for two different amounts. Keep your envelope: the money is the same either way.

Why does the 1.25x argument feel so convincing?

Because each step looks correct on its own. Doubling and halving are both plausible, the probabilities are both one half, and the arithmetic is right. The error is in the setup: the symbol x changes meaning between the two branches, and nothing flags that.

Does opening the envelope change the answer?

It can, but only if you also know the range of amounts. Opening the envelope tells you a number. Without a range, that number says nothing about whether you hold the smaller or the larger amount. With a range, a very large number means you probably hold the larger envelope, so keep it.

Source: the question is reported as a Graduate Trader interview question at Vivienne Court in a Glassdoor interview report, and it appears in standard quant interview collections. The solution and all text here are original.