(in attempting to answer the original question .... )
One way to estimate a daily VaR at the 95% confidence level, for the system/portfolio (using historic simulation) would be to calculate the daily changes for the system for maybe the past 125 trading days (~6 trading months), sort these lowest to highest, take the negative value which reperesents that level of loss where 95% of daily losses are less (i.e. smaller losses) than this, and use that as the VaR estimate.
Any other time period could be used, including the entire history available. Daily changes as defined above could be in percent of trading level or absolute currency values, depending on the desired output. The effects of compounding will tend to skew the accuracy of the estimate when using absolute currency values and this skew becomes greater the longer the lookback period. A percent of trading level (or account value) value tends to avoid this skew. This method of estimating VaR will usually produce estimates that are close to more computationally intensive estimators. There is no 'correct' estimate to my knowledge, since it is an estimated value with a confidence level to begin with.
Hope this helps.
VaR
Although I'm not a big fan of VaR (I certainly wouldn't rely on it for my own trading) it is often quoted in the big, bad world of hedge funds including those that use LTTF.
With this in mind, is it possible to get an estimation of % equity risked (or AUM) if I just had monthly and yearly returns for a given level of VaR ?
With this in mind, is it possible to get an estimation of % equity risked (or AUM) if I just had monthly and yearly returns for a given level of VaR ?