Skip to content

Add proper walk-forward cross-validation to training pipeline - #189

Open
ch55secake with Copilot wants to merge 4 commits into
mainfrom
copilot/add-walk-forward-validation
Open

Add proper walk-forward cross-validation to training pipeline#189
ch55secake with Copilot wants to merge 4 commits into
mainfrom
copilot/add-walk-forward-validation

Conversation

Copilot AI commented Apr 19, 2026

Copy link
Copy Markdown
Contributor
  • Understand codebase and existing WalkForwardValidator in src/experimental/
  • Enhance WalkForwardValidator to support expanding/sliding window types, per-fold R²/RMSE/directional accuracy/Sharpe metrics, and degradation flagging
  • Add walk_forward_validate() method and constructor parameters to StackedModelTrainingPipeline
  • Wire walk-forward validation into train() when enabled
  • Update src/experimental/__init__.py exports
  • Add 28 tests for walk-forward validation (split, metrics, validate, pipeline integration)
  • Run linter and tests (459 passed, 0 failures)
  • Address code review: fix unused params in test factory, add comment to fallback wf_results

Copilot AI and others added 2 commits April 20, 2026 09:26
- Enhance WalkForwardValidator with expanding/sliding window modes,
  per-fold R²/RMSE/MAE/directional-accuracy/Sharpe metrics, and
  regime-sensitivity degradation detection
- Add walk_forward_validate() method to StackedModelTrainingPipeline
  with configurable window type, train/test windows, and retrain freq
- Wire walk-forward validation into train() when use_walk_forward=True
- Export WindowType from src/experimental/__init__.py
- Add 28 tests covering split logic, metrics, validate(), and pipeline
  integration for both expanding and sliding window modes"

Agent-Logs-Url: https://github.com/ch55secake/hyperion/sessions/c3985177-92c0-4ab8-93ff-0102e83df3ef

Co-authored-by: ch55secake <87881861+ch55secake@users.noreply.github.com>
@ch55secake
ch55secake marked this pull request as ready for review April 21, 2026 00:06
@ch55secake ch55secake changed the title [WIP] Add proper walk-forward cross-validation to training pipeline Add proper walk-forward cross-validation to training pipeline Apr 21, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

ML: Add proper walk-forward cross-validation to the training pipeline

2 participants