Classification Metrics Quant Interview Guide
Classification metrics quant interview guide for accuracy, precision, recall, false positives, class imbalance, examples, and metric choice.
Candidates evaluating binary classifiers, signals, and screens.
Metrics depend on the decision
Classification metrics summarize different errors and successes. The right metric depends on whether false positives, false negatives, ranking quality, or calibration matters most for the decision.
Accuracy can be misleading
When classes are imbalanced, a model can have high accuracy by mostly predicting the common class. Interviewers often expect you to notice the base-rate problem before celebrating accuracy.
Concrete example
If only 1 percent of cases are positive, a model that always predicts negative is 99 percent accurate but useless for finding positives. Precision and recall may be more informative.
Tie metrics to cost
A trading screen, fraud detector, and hiring filter can have different error costs. Explain which error is more expensive in the prompt before choosing a metric.
Common mistakes
Candidates often list metrics without choosing. A strong interview answer selects a metric, justifies it, and names what it fails to capture.
Practice the pattern
Use the LeetQuidity curriculum and calibration to turn this topic into a focused practice plan.