Page 1 of 1

chaos detector

Posted: Sat Jun 03, 2006 3:41 pm
by Quantus
Hello,

are there any quants/techs around who maybe know or have (or know where to get :roll: ) tools for filtering financial (tick!) data (best: in real-time) to determine if the actual structures are chaotic, noisy (different types of noise), deterministic... e.g. telling you the lyapunov exponent with not too many points for analysis...

Does anybody have experience f.e. with the "Chaos Toolkit" from Scientio (http://www.scientio.com/chaoskit.aspx)?

Unfort. I'm not an expert at all in these fields, but I guess this could be a very important piece of the "puzzle" - trading only in non-chaotic structures (I know there are many different definitions)... just one way to filter, others maybe use the equity curve in combination with technical indicators to decide if to trade the system(s) or not..

However, all hints/help/discussion concerning a "chaos detector" is highly appreciated!

Quantus


p.s.: I like the flow for a `proposed hybrid system modelĀ“ from Dr. Clarence Tan in his article "A Hybrid Financial Trading System Incorporating Chaos Theory.. ", page 6:

www.smartquant.com/references/NeuralNet ... ural29.pdf

Posted: Mon Jun 19, 2006 8:42 pm
by eudamonia
Quantus,

Well I'm not an expert, but I do have a similar interest. The "Chaos Toolkit" looks interesting and for the price they are charging I would guess you ought to be able to get more information than is on the website. Some examples of practical application are a definate must. I've never heard of these folks but they look new.

Also, I'm guessing by the type of math terms that the "Chaos Toolkit" folks are referring to that their system is not an ANN or a GA. It's closer to "fuzzy logic" would be my guess. Safir-X has a similar program that detects patterns out of raw finacial data using fuzzy logic, but it is heinously expensive IMHO. http://www.sirtrade.com/compare.htm

Edward

Posted: Wed Nov 17, 2010 4:44 pm
by xtremeforex
Don't over-complicate trading. This is all great for research purposes but I can tell you that predicting the future is impossible and any attempt to do so may actually work but your information will always be lagging and delayed, therefore it won't help for making your next move or determining risk sizes. I've been through this route of chaos theory, fuzzy logic etc etc and whilst successful all the wonderful predictions worked but were always delayed. Delayed information is useless, by the time your algorithms do their computations, it's too late. This is the inherent problem.

If you're going to do any mathematical/physical based research in this area , perhaps focusing on effective trade/managemet would be time well spent. Believe me, you can get into 3D analysis with trade/risk management if you're creative..