statBootstrap: Stationary bootstrap method

In skoestlmeier/monotonicity: Test for Monotonicity in Expected Asset Returns, Sorted by Portfolios

Description

statBootstrap implements the stationary bootstrap method from Politis & Romano (1994, JASA). This function generates bootstrap samples of the matrix data and returns the time indices for each sample.

Usage

statBootstrap(T, bootstrapRep = 1000, block_length)

Arguments

T	A scalar, the number of time series observations to generate.
bootstrapRep	A numeric scalar: the number of used bootstrap samples.
block_length	A numeric scalar: The average length of the block to usefor the stationary bootstrap. This parameter is related to how much serial correlation is in your data. Use 10/6/3/2 as the block length if data is measured in daily/monthly/quarterly/annual returns.

Value

statBootstrap returns an object of class "matrix":

"matrix":

A "T x bootstrapRep" matrix of time indices for each bootstrap sample.

References

Politis, Dimitris N. & Romano, Joseph P. (1994): The Stationary Bootstrap. Journal of The American Statistical Association, 89, No. 428, p. 1303–1313. doi: 10.1080/01621459.1994.10476870.

Examples

## Assuming daily return data for 100 time series observations.
## The returning matrix for default settings contains 1,000 bootstrap samples.
bootstrap_sample <- statBootstrap(T = 100, block_length = 10)

## 200 bootstrap samples using monthly return data with 250 time series observations.
bootstrap_sample <- statBootstrap(T = 250, bootstrapRep = 200, block_length = 6)

skoestlmeier/monotonicity documentation built on Dec. 29, 2021, 9:54 a.m.