As the holding period increases, the number of observations decreases. This is a practical problem encountered when estimating historical simulation ES or VaR. The reason behind this is that as we increase the holding period, we are essentially aggregating more data points into a single observation. For instance, if we have 1000 observations of daily Profit/Loss (P/L), that corresponds to four years’ worth of data at 250 trading days a year. If we decide to use a weekly holding period instead of a daily one, each weekly P/L will be the sum of five daily P/Ls, reducing our total number of observations to 200. If we further increase the holding period to a month, each monthly P/L will be the sum of 20 daily P/Ls, reducing our total number of observations to just 50. Therefore, as the holding period increases, the number of effective observations rapidly falls, imposing a major constraint on how large the holding period can practically be when estimating historical simulation ES or VaR.

Choice A is incorrect. The number of observations does not decrease as the holding period decreases. In fact, it's the opposite — a shorter holding period would typically result in more observations.