The correct answer is A.

To calculate the Value at Risk (VaR) using historical simulation at a 90% confidence level, we must identify the threshold where exactly 10% of the return distribution falls in the tail.

Given there are 30 observations, we calculate the number of observations in the tail (representing the worst losses): Number of tail observations = Total Observations × (1 - Confidence Level) Number of tail observations = 30 × 0.10 = 3

The worst returns sorted from worst to best are:

1. -20%
2. -18%
3. -18%
4. -17%
5. -15% ...

To have exactly 3 returns worse than the VaR threshold, we look at the 4th worst observation, which is -17%. Thus, the returns strictly worse than -17% are exactly the 3 worst returns (-20%, -18%, and -18%), meaning there is exactly a 10% chance of experiencing a loss greater than 17%. As VaR is typically expressed as a positive number denoting the loss amount, the VaR at the 90% confidence level is 17%.