QM006 Hands-on Lab

See what IID resampling destroys when observations are serially dependent

Goal

This short exercise compares IID resampling with a moving-block bootstrap on the same synthetic dependent series.

Start here

Download the complete lab bundle, or use the direct Python and R scripts.

Baseline

python labs/python/qm006_hands_on.py

Then try:

python labs/python/qm006_hands_on.py --block-length 4
python labs/python/qm006_hands_on.py --block-length 8
python labs/python/qm006_hands_on.py --block-length 32

Interpretation

The exercise is designed to show that resampling individual observations can remove serial dependence that matters for uncertainty. The numerical gap is specific to this synthetic series; block length is a tuning choice, not a universal constant.