Cross-Validation — bpleone / trade

K-fold CV splits the training data into K parts. The model trains on K-1 parts and tests on the held-out part — repeated K times so every sample gets tested. Average accuracy is a robust estimate. Cross-validated accuracy ≠ in-sample accuracy. If they differ a lot, the model is overfitting.

Avg CV Accuracy

Std Deviation

Across folds

In-Sample Acc

Trained on all data

Overfitting Gap

📋 Per-Fold Results

FOLD

BAR

ACCURACY

PRECISION

RECALL

📊 Accuracy Distribution Across Folds