Compute standard error on the mean of a timeseries of (possibly correlated) measurements. This is technique is superior to standard methods based on MSDs because it correctly considers the degree to which your data is correlated
import numpy as np
from flyvbjerg_petersen import fp_stderr
data = np.random.randn(1000)
err = fp_stderr(data)