Model VaR is predictive machine to make many trees of possible events outcomes. Application can complete this data into the single value and show it's so easy. Each single tree it's probabilities to get various money income (positive or negative) from any shares with specific money amount to the invest ... Recalculate each day after market' close.
Undefined peoples criticized VaR and say that it's important cause of 2008 year crisis. Author say about:
- That this model was working by the historical data between 1980 and 2005 years. That isn't guaranties that it's the best predict for analyze;
- Better to use data from the large historical period;
- 1% to chance to lost 500 millions is only chance, it isn't guarantee;
- It's never not fully information about risk and profit ... ;
- In't copy of "Beer test" but it's just with opposite results;
Possible critical mistakes
- Assuming events are independent events when they are not! Need to ask, "Is it possible to be together with specific conditions?". Write answer also to test it;
- Not understanding when event ARE independent! Write ask and answer also to test it;
- Clusters happened. Many equal events in the one place, but without communications.
- Context is important. It can change probability value;
- In the long time value tends to the average value;
- Do need to divide general observation of people to male and female groups?