In a scenario involving a company's 1-day 99.5% Value at Risk (VaR) model over a period of 10 years, with a 95% confidence level, and assuming 250 trading days per year and that daily returns are independent and identically distributed (i.i.d.), what is the approximate maximum number of days in which the daily losses exceed the 1-day 99.5% VaR, yet still imply that the model is adequately calibrated?