July 16th, 2012, 11:03 pm
It cannot be done with the data you have, it would only be misleading.The best approach I can think of (and it's not very good) is to look at the losses from the fund categories during bad periods, adjust for the volatility of your funds, and add some idiosyncratic risk.For example, if an index of merger arb funds over 20 years has a monthly VaR of 3% and your merger arb fund has a monthly vol that is 80% of the index, you might guess a 2.4% VaR, plus add something for the idiosyncratic risk.For a portfolio, look at the returns for a vol and type matched portfolio of the index, and again add for idiosyncratic risk.That is better than trying to use your short historical data.