# ES using historic simulation

Why is the data obtained from 91-100 days all eliminated from the calculation of the 1-day 95% ES? My interpretation is because the first day to calculate the 95% ES should be the 90th day? But I can't quite convince myself, could someone explain to me?

• I think this is what is going on:<br> The data in the table was for the last 100 days as of time T. But now it is T+10 ("ten trading days have elapsed"). So we have to rebuild the table of 100 recent days: data for days 91,92,...100 drops out (too old, no longer within 100 days of T+10) and we may have to add new data. <br> But we are told: "over the last 10 days the worst return was -2590". So none of the recent returns are below the VaR of 3700, so there is nothing to add. <br> We then calculate the ES over the non-deleled entries in the table. Mar 13, 2020 at 17:33
• (But I thought 95% means all but 1 out of 20, so why are they looking at a list of 10 entries and not 20 entries? A mystery.) Mar 13, 2020 at 17:47