Skip to content

Package: regimeflow::walkforward

Summary

Walk-forward optimization and parameter search. Provides optimizers, sampling distributions, and trial bookkeeping.

Related diagrams: - Walk-Forward

File Index

File Purpose
regimeflow/walkforward/optimizer.h Walk-forward optimizer and trial control.

Type Index

Type Description
WalkForwardOptimizer Runs train/validate/test splits.
WalkForwardConfig Optimization configuration.
ParameterDef Parameter definition for optimization.
ParameterSet Parameter name/value map.
WindowResult Single window outcome.
WalkForwardResults Aggregated walk-forward outputs.

Lifecycle & Usage Notes

  • Optimizers integrate with BacktestRunner to execute trial batches.
  • Results can be exported through ReportWriter in metrics.

Type Details

WalkForwardOptimizer

Coordinates train/validate/test splits and aggregates trial outcomes.

Methods:

Method Description
WalkForwardOptimizer(config) Construct optimizer.
optimize(params, strategy_factory, data_source, full_range, detector_factory) Run walk-forward optimization.
on_window_complete(callback) Register window-complete callback.
on_trial_complete(callback) Register trial-complete callback.
on_regime_train(callback) Register pre-regime-train hook.
on_regime_trained(callback) Register post-regime-train callback.
cancel() Cancel running optimization.
### WalkForwardConfig / ParameterDef / ParameterSet

Configuration and parameter definitions for optimization.

Fields:

Member Description
WalkForwardConfig Optimization configuration fields and enums.
ParameterDef Name/type/range/distribution.
ParameterSet Parameter name/value map.

ParameterDef

Parameter definition for optimization.

Distribution

Sampling distribution enum.

OptMethod

Optimization method enum.

WindowType / OptMethod / Distribution

Optimization enums for windowing and sampling.

RegimeTrainingContext

Context passed to regime training callbacks.

TrialOutcome

Result of a single optimization trial.

WindowResult / WalkForwardResults

Per-window and aggregated optimization outputs.

WalkForwardResults

Aggregated walk-forward results.

Method Details

Type Hints:

  • params → std::vector<ParameterDef>
  • strategy_factory → std::function<std::unique_ptr<strategy::Strategy>(const ParameterSet&)>
  • data_source → data::DataSource*
  • full_range → TimeRange

WalkForwardOptimizer

optimize(params, strategy_factory, data_source, full_range, detector_factory)

Parameters: parameters, strategy factory, data source, range, optional detector factory. Returns: WalkForwardResults. Throws: None.

on_window_complete(callback) / on_trial_complete(callback)

Parameters: callbacks. Returns: void. Throws: None.

on_regime_train(callback) / on_regime_trained(callback)

Parameters: callbacks. Returns: void. Throws: None.

cancel()

Parameters: None. Returns: void. Throws: None.

Usage Examples

#include "regimeflow/walkforward/optimizer.h"

regimeflow::walkforward::WalkForwardConfig cfg;
regimeflow::walkforward::WalkForwardOptimizer optimizer(cfg);

See Also