Skip to main content

Table 2 Comparison of unbiased overlapping return standard deviation and skewness estimators as a function of weighting scheme, sample size n, and overlapping return period q

From: Unbiased weighted variance and skewness estimators for overlapping returns

 

n=256

 

n=512

 

n=1024

 
 

q=5

q=21

q=63

q=5

q=21

q=63

q=5

q=21

q=63

Std(%)

         

U

1.764

3.146

4.086

1.885

3.459

5.225

1.729

3.130

4.387

BT

2.167

3.810

4.359

1.999

3.519

5.330

1.611

2.847

3.845

PR

2.409

4.321

4.679

1.894

3.196

4.963

1.512

2.634

3.260

TK

2.215

3.892

4.388

1.997

3.480

5.328

1.556

2.730

3.501

QS

2.038

3.561

4.246

2.001

3.555

5.380

1.648

2.930

3.978

Skewness

         

U

− 0.489

− 0.296

0.356

− 1.038

− 0.356

− 0.467

− 0.858

− 0.400

− 0.866

BT

− 0.612

− 0.525

0.299

− 1.200

− 0.310

− 0.343

− 0.748

− 0.401

− 0.985

PR

− 0.582

− 0.752

0.294

− 1.355

− 0.387

− 0.668

− 0.508

− 0.335

− 0.488

TK

− 0.616

− 0.623

0.334

− 1.280

− 0.308

− 0.316

− 0.610

− 0.328

− 0.801

QS

− 0.601

− 0.446

0.318

− 1.184

− 0.298

− 0.318

− 0.791

− 0.401

− 0.986

  1. Here, we use the following abbreviations for different kernel based weighting schemes: U uniform, BT Bartlett, PR Parzen, TK Tukey-Hanning, and QS Quadratic Spectral