WalkForwardOptimizer

WalkForwardOptimizer

public WalkForwardOptimizer(BruteForceOptimizer optimizer)

walkForwardOptimizer = WalkForwardOptimizer(optimizer)

Walk-forward optimization: the parameters are searched on one stretch of history and the best of them tested on the stretch after it, window by window, so every out-of-sample result comes from values chosen without seeing it.

optimizer

The search run on each in-sample stretch, and the runner of each out-of-sample one.

RunAsync

public IAsyncEnumerable<WalkForwardStep> RunAsync(Strategy strategy, IStrategyParam[] parameters, IEnumerable<WalkForwardWindow> windows, Func<Strategy, decimal> fitness, CancellationToken cancellationToken)

result = walkForwardOptimizer.RunAsync(strategy, parameters, windows, fitness, cancellationToken)

Runs the windows in order.

strategy

The strategy whose copies are run.

parameters

The parameters to search, with their optimization ranges.

windows

The windows to run.

fitness

Scores a strategy after its in-sample run; the highest score is chosen.

cancellationToken

CancellationToken

Rückgabe: One step per window that had anything to choose from.