Search All of the Math Forum:
Views expressed in these public forums are not endorsed by
NCTM or The Math Forum.


networm
Posts:
327
Registered:
10/6/05


Possible to estimate a unknown barrier from only one realization of time series?
Posted:
Jul 12, 2006 2:41 AM


Suppose I model an observed process Ct as a discretized Geometric Brownian Motion process.
And there is a unknown but fixed/constant barrier, initially below the process Ct.
If the process Ct hits the barrier, Ct will become zero ever since.
I observe the time series up until time t(the current time), and I know at each sample point, the process did not hit the unknown barrier yet.
Can I estimate the three parameters (mu, sigma, barrier) from this one realization of the timesereis? (mu and sigma are for the GBM. )



