test_acf_bootstrap

Notedocblock

test_acf_bootstrap(data, n_boot, h; alpha=0.05, block_size=1)

Compute a bootstrap hypothesis test for the classical autocorrelation at lag h and return an ACFTestResultBoot with the bootstrap critical value, p-value, and reject decision.

Unlike test_acf(), this does not rely on the asymptotic N(0, 1/n) distribution and is therefore more reliable for short time series or non-Gaussian data.

• data: the time series.
• n_boot: number of bootstrap replications.
• h: lag.
• alpha: significance level (default 0.05).
• block_size: set > 1 for a block bootstrap that preserves serial dependencies.