Independent increments of a Gaussian martingale Created 2026-09-24 Updated 2026-11-05
A centered continuous Gaussian process that is also a martingale has independent increments. Every future increment has zero covariance with every finite vector of past values by the martingale property; uncorrelated jointly normal variables are independent.
Let . Under the usual nondegenerate regression assumption , maximizing the normal likelihood is equivalent to minimizing the residual sum of squares. The normal equations and give the maximum-likelihood estimators
Write and . Then , , and the residual is . Independence of the errors gives
For each ,
Thus the three requested variables are pairwise uncorrelated.
They are linear functions of the jointly normal error vector. More strongly, the whole vector has a multivariate normal distribution, and its block covariances show that , and the residual vector are independent, by uncorrelated jointly normal variables are independent. Independence is preserved when applying a measurable function to the residual block. Hence
The residual projection is , of rank , so additionally . This is the orthogonal projection of a Gaussian vector interpretation of the same result. The printed design conditions alone allow ; then the fit is exact, the positive-variance likelihood has no finite maximizer, and the residual test in part (b) is undefined. The standard testing question therefore needs .
Let
They are jointly normal because they are linear combinations of a Gaussian vector. When ,
Hence uncorrelated jointly normal variables are independent. Their means and variances are
and
Therefore
independently.
Fix . Since is a martingale,
Every finite vector consisting of and past values has a multivariate normal distribution. Therefore uncorrelated jointly normal variables are independent, so the increment is independent of every finite vector of past values. A Monotone class theorem then extends this to independence from . This is the independent increments of a Gaussian martingale.
Uncorrelated random variables Created 2026-09-24 Updated 2026-10-14
For an explicit counterexample, let take , , with equal probabilities, and let . Then , so . Nevertheless, is completely determined by : , while . Thus these are uncorrelated random variables with statistical dependence. In this example, opposite contributions from the odd function cancel; the even function still retains information about . The last panel of the covariance diagram shows the relationship.
There are useful special cases. Uncorrelated jointly normal variables are independent, and the Bernoulli independence from zero covariance result supplies another setting where the converse holds. Their distribution assumptions do the extra work; zero covariance alone does not.