In VaR calculations using order statistics, confidence intervals are often derived using the binomial distribution, as VaR is typically based on quantiles of the return distribution. The discrete nature of the binomial distribution makes it challenging to calculate precise confidence intervals because it introduces granularity in the estimation process. This can result in confidence intervals that are less precise or overly wide, depending on the sample size and confidence level.

A is incorrect. Order statistics, as a method, does not directly address time-varying volatility. This is a limitation of the underlying historical simulation approach upon which order statistics is based, but it's not a challenge inherent in the order statistics method itself. The challenge is in the data it uses.

B is incorrect. Order statistics assume independence, which is a limitation of the method, not a challenge inherent in the calculation itself.

C is incorrect. Order statistics are a non-parametric method and do not require complex distributional assumptions.