All Things Statistical - page 9
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
Regime Detection
Good Morning All,
Posting something i read years ago (as are most posts on this thread) that was inspiring to me when i read it. Regime Detection illustrations below based on simulated data. The final post/study here based on real-world data naturally concludes that reality is way way way more difficult/tougher. But good as an example of showing what one can do with regime-modelling i guess if one can successfully to overcome the multitude of real-world issues.
"The first 300 observations were used to calibrate this model, the next 300 observations were used to see how the model can describe the new infromation. This model does relatively well in our toy example"
http://jonathankinlay.com/index.php/2011/04/regime-switching-market-state-modeling/
https://systematicinvestor.wordpress.com/2012/11/01/regime-detection/
https://systematicinvestor.wordpress.com/2012/11/15/regime-detection-pitfalls/
Wintersky
Cheers
Statistical Testing Of Demark Indicators
A Paper Testing 3 of Tom Demark's indicators for those who might be interested:
http://www.er.ethz.ch/publications/LISSANDRIN_demark_thesis_final.pdf
There has been a totally mixed review by people who used his indicators, which some suggests that the reason might be due to non-universal applicability in different markets ie. works well for some but not others due to different market characteristics. Good luck
Wintersky
Cheers
Can One Beat A Random Walk?
Intelligent Trading: Can one beat a Random Walk-- IMPOSSIBLE (you say?)
Wintersky
Cheers
This paper introduces an extension to the method, which computes the correlation dimension of Japanese candlesticks instead of correlation dimension of commonly used ‘close’ points of the economic time series, the results show lower chaotic dimension which implies a less complex dynamical system behind the tested macro-economical time series, a lower dimension in a dynamical system is satisfying.http://waset.org/publications/8276/estimating-correlation-dimension-on-japanese-candlestick-application-to-forex-time-series
This paper introduces an extension to the method, which computes the correlation dimension of Japanese candlesticks instead of correlation dimension.....
Good Morning All,
Interesting. The only thing here is that this paper doesnt actually validate the wide variety of different sets of candlestick formations as it seems to study single candlesticks serially in terms of correlation dimension. More like a study of single bar candlesticks and a validation of additional information content leading to a possible edge that can be derived from the relationship of OHLC. A nice paper indeed.
Pretty sure alot of people are well-acquainted with the site below, but still a link for clustering to find C-patterns. One weird thing though is that from my memory, a hammer is supposed to be followed by opposite directional price movements but in his example, it was the reverse..... a hammer followed by down move???
Intelligent Trading: Quantitative Candlestick Pattern Recognition (HMM, Baum Welch, and all that)
Wintersky
Cheers
Effective Sample Size & Autocorrelations
Good Day All,
This is killing me ****** LOL@@@ :
"These examples show that our rather vague intuitive feeling that “positive correlation tends to decrease information content in an experiment” is very far from the truth, even for rather simple normal experiments with three observations"
https://golem.ph.utexas.edu/category/2014/12/effective_sample_size.html
How to use the Autocorreation Function (ACF)? | CoolStatsBlog
Wintersky
Cheers
Autocorrelations...
Good Day All,
Still in the midst of investigating Autocorrelations....
A funny idea here is to not use it as per the original intention but as a measure of "smoothness" instead... lawls
How to measure smoothness of a time series in R? - Cross Validated
Some Stylized Facts for Autocorrelation:
martinsewell.com: Finance: Stylized Facts: Dependence
It's also clear that Autocorrelations is all over the place for nonstationary time series... as evidenced by the wide range of coefficients in different timesteps as seen in the below picture... Adjusting for AR(1) (as a minimum measure) creates less Type 1 Errors but increases Type 2 Errors.... but the possibility of doing anything more stretches into the realm of prediction perhaps.... maybe the solution is to just increase sample size and leave as be even without accounting for AR(1)?
http://faculty.haas.berkeley.edu/lyons/Eom%20partial%20price%20adjustment.pdf
Wintersky
Cheers
Big funds borrow lot of ideas from bio-statistics , here is an award winnig paper : Exploratory data analysis with noisy measurements Exploratory data analysis with noisy measurements - Wentzell - 2012 - Journal of Chemometrics - Wiley Online Library
Big funds borrow lot of ideas from bio-statistics , here is an award winnig paper : Exploratory data analysis with noisy measurements Exploratory data analysis with noisy measurements - Wentzell - 2012 - Journal of Chemometrics - Wiley Online Library
Nice. Thanks
Big funds borrow lot of ideas from bio-statistics , here is an award winnig paper : Exploratory data analysis with noisy measurements Exploratory data analysis with noisy measurements - Wentzell - 2012 - Journal of Chemometrics - Wiley Online Library
Looks cool. Thanks for the link. Will be reading to absorb it. Reading too many things a the moment. Currently looking into volatility estimation issues...
Wintersky