Struct WalkForwardOptimizer 

pub struct WalkForwardOptimizer { /* private fields */ }

Walk-forward optimizer: splits bars into rolling train/test windows, runs a grid search on training data, and evaluates the winning parameters on held-out test data.

The strategy factory closure receives the training bar slice and the candidate parameter HashMap<String, f64>. It should return a Box<dyn Strategy> initialized with those parameters.

Example

See the module-level documentation for a complete example.

Implementations

impl WalkForwardOptimizer

pub fn new( config: WalkForwardConfig, bt_config: BacktestConfig, ) -> Result<Self, FinError>

Constructs a new optimizer.

Errors

Returns FinError::InvalidInput if the config is invalid.

pub fn run<F>( &self, bars: &[OhlcvBar], make_strategy: F, ) -> Result<WalkForwardResult, FinError>

where F: FnMut(&[OhlcvBar], &HashMap<String, f64>) -> Box<dyn Strategy>,

Runs the walk-forward optimization.

make_strategy is called once per (parameter combination × window). It receives:

• train_bars: the training slice for fitting
• params: a HashMap<String, f64> with the current grid point

It must return a Box<dyn Strategy> configured with params.

Errors

• FinError::InvalidInput if bars is too short for one window.
• Propagates any FinError from individual backtest runs.

pub fn config(&self) -> &WalkForwardConfig

Returns the configuration.

pub fn bt_config(&self) -> &BacktestConfig

Returns the backtest configuration.