See article on Long Memory and Regime Shifts in Asset Volatility

http://jonathankinlay.com/index.php/2011/03/long-memory-and-regime-shifts-in-asset-volatility/

"The article goes on to review some of the statistical procedures that have been developed to detect regime shifts, due to Bai (1997), Bai and Perron (1998) and the Iterative Cumulative Sums of Squares methodology due to Aggarwal, Inclan and Leal (1999). The article illustrates how the ICSS technique accurately identifies two changes of regimes in a synthetic GBM process."

Does anybody have links to MQL code which implements the basics of this approach?

Another idea but more refined than CUMSUM....For Joe Ellsworth....EXCEL formulas are in the article....http://www.biopharminternational.com/use-change-point-analysis-process-monitoring-and-control

Folks, LOTS of more refined ideas are in the bio-engineering field....

Financial article with application of change point idea

http://personal.lse.ac.uk/schroeam/ATE_15_July_2013.pdf

Thank you for the links. The second paper is very interesting.

The first paper is an interesting development of CUSUM as well. Though note that it oversimplifies CUSUM, at least as used in more sophisticated process control. CUSUM incorporates the likelihood function (w in http://en.wikipedia.org/wiki/CUSUM) that can be used to control for Type I errors and manage the confidence level.

Hi Greg,

The point of posting the papers was to introduce you to the multiple ways of doing change point analysis. Clealry, the first article is not as informative, since it is not intended to be up for academic publication.

In the BioPharm International report summary,

"Using CUSUM charts in combination with bootstrapping, change-point analysis can produce a confidence level for every change in the mean or variation detected."

**********

http://www.biopharminternational.com/use-change-point-analysis-process-monitoring-and-control?id=&sk=&date=&%0A%09%09%09&pageID=5

***********

Essentially, this report combines bootstrapping with CUMSUM..1 form of Changepoint analysis

The second paper is addressing change points via binary segmentation and wavelets

***************

http://personal.lse.ac.uk/schroeam/ATE_15_July_2013.pdf

***************

You can see each of these methods and more presented at

The Changepoint Repository

************

http://www.changepoint.info/introduction.html

*************

Alexander Gray at the Georgia Insitute of Technology gives some technical descriptions and explanations of the ideas in some of his classes.

Here is another ChangePoint article utilizing CumSum.

Best,

John

*************

http://www.variation.com/cpa/tech/changepoint.html

*****************

I agree the 2nd paper is quite interesting.

Wow! A few real gems here. Thanks especially for passing along The Changepoint Repository.

Before anyone becomes too slap happy about the second paper, realize that this approach has been around for a while. Note, the binary segmentation is the familiar bisection algorithm from numerical analysis. Utilizing wavelets for time series analysis has been around for quite a while. I started using both in 2001.

I have monitored managers using "CUSUM" which has its roots based in signal processing and was used by IBM in their pension fund monitoring back in the 60's. A great place to look for papers, including calculations, and software is Northfield (northinfo<.>com - its President is Dan diBartolomeo. The firms I worked for which were eventually sold up and finally ended up in RogersCasey's hands used it. So if you've come across monitoring material from them, there are many similarities. Also, I believe it was Sapient that had a seminar that included a spreadsheet last year. Good Luck. Contact me if you wish to discuss further. Cheers,

Thank you, Gerald. Indeed, I began investigating CUSUM after coming across it in a Northfield presentation. The more specific problem is: After a statistically significant signal/regime change is detected, ranking the magnitude of the change and comparing these across several time series.

Another thread brought up this helpful presentation http://qwafafew.org/images/uploads/boston/dan_cusum_2005.pdf that touches on the problem.

Another performance metric is the Omega ratio. Here are some presentations from the folks at Fuqua.

%%

https://faculty.fuqua.duke.edu/~charvey/Teaching/BA453_2005/Keating_An_introduction_to.pdf

%%

https://faculty.fuqua.duke.edu/~charvey/Teaching/BA453_2006/Omega_SlideShow.pdf

%%

https://faculty.fuqua.duke.edu/~charvey/Teaching/BA453_2005/Keating_The_omega_function.pdf

%%

https://faculty.fuqua.duke.edu/~charvey/Teaching/BA453_2005/Keating_A_universal_performance.pdf