statistics

today,

i studied about statistics

especially about a modelling method of market.

currently most used method for approximating method for the market is VaR,

but it cannot assume that the market value is not on the normalized distribution,

because the actual values are fat-tail and asymmetry.

so i studied about the autoregressive conditional heteroskedasticity.

( toooooooo long word... )

it assume the error of the market value as the normalized distribution,

but assume that the standard deviation is not fixed value,

but changed as the time sequence changes.


and, i researched about this method...

but i couldn't understand how to estimate the actual value from past values.

i can get the estimated standard deviation,

but the return is calculated with the standard deviation and error,

which is on the normalized distribution.

but, there are no description about calculating the error.

how can i calculate...

i have to research it until Monday.

mmmh, can i sleep this weekend?...