1
$\begingroup$

Way back when, I used to work in finance, and I remember helping a coworker use some kind of block bootstrap. (I believe the application was: we had weekly data on some financial indicator X, along with weekly data on some stock, and we wanted to measure how well X could be used to predict the stock's movements. And I believe we needed to bring in the bootstrap, because we only had a couple months of weekly data, so we didn't really have many datapoints. I might be misremembering all this, though.)

In any case, I totally forget now how the block bootstrap worked, and I want to remember/review/learn more, so can anyone suggest a good tutorial on it? I tried googling, but all I found were some random research papers. I also tried looking in my copy of Efron & Tibshirani's "An Introduction to the Bootstrap Book", but didn't find anything (unless it's under a name other than "block bootstrap").

Improve this question
$\endgroup$
  • $\begingroup$ i have found some useful content in Hardle, Horowitz et al But the methods mentioned there mainly deal with estimating mean (or, in general, moments) of time series data.....i wanted to know if the method applies to regression set-up also.I am currently knee-deep in a live project involving this without much idea of how to proceed....Any help would be greatly appreciated...thanks $\endgroup$ – user24466 Apr 17 '13 at 11:05
  • $\begingroup$ Do you want to give a full reference to Hardle, Horowitz et al.? The second part of your answer is actually a new question. Since this is not a discussion list, you should post it as a new question. $\endgroup$ – Maarten Buis Apr 17 '13 at 11:47
4
$\begingroup$

I have relied on Resampling Methods for Dependent Data by S.N. Lahiri and found it quite helpful. Once you determine some flavors you want to look at more closely (e.g. Circular Block Bootstrap, Stationary Block Bootstrap) it will be easier to find on-line resources discussing actual use cases and implementation details.

Improve this answer
$\endgroup$
3
$\begingroup$

Try the Handbook of Computational Statistics, Part III, section 2.4.

Improve this answer
$\endgroup$
  • 1
    $\begingroup$ The url had changed. I've now fixed it. $\endgroup$ – Rob Hyndman Apr 19 '13 at 0:22
1
$\begingroup$

The textbook by Shumway and Stoffer has a short section on bootstrapping time series (state-space) models. Also you may look to:

Pfeffermann, D. and Tiller, R. (2005) Bootstrap approximation to prediction MSE for State-Space models with estimated parameters, Journal of Time Series Analisys, 26, 893-916,

and references therein.

Improve this answer
$\endgroup$
1
$\begingroup$

For an evaluation of its effectiveness and a comparison to alternative methods, see:

Bertrand, Marianne, Esther Duflo and Sendhil Mullainathan. 2004. "How Much Should we Trust Difference-in-Difference Estimates?" Quarterly Journal of Economics. 119 (1): 249-275. [pre-publication version]

Improve this answer
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.