How does SVM behave under extreme class overlap?

SVM cannot perfectly separate highly overlapping classes and relies on soft margins.

When distributions overlap heavily, SVM trades off margin maximization and misclassification using slack variables. Performance depends on feature representation quality.