I am studying two stochastic timeseries $X(t)$ and $Y(t)$, and I have reasons to believe that they are related by the following equation

$X(t) = Y(t-\Delta t) + dW(t)$

where $dW$ is the increment of a Wiener process, i.e. white gaussian additive noise, with a certain variance $\sigma^2$. I am trying to estimate $\Delta t$. In this case $\sigma^2$ is known, the system is stationary and ergodic, and the equation applies to detrended $X,Y$. Otherwise, I can also assume that the expected values are zero, i.e. $E(X)=E(Y)=0$.

I have some timeseries of $X,Y$ at my disposal, and have attempted the problem by estimating the covariance $C_{X,Y}(t)$ of $X$ and $Y$, but failed to observe a drop in the covariance $C_{X,Y}(\Delta t)$ as I was expecting.

Are there any established techniques to estimate the delay $\Delta t$? This seems to me a rather common problem, and I am struggling to find good references on this.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.