When the hurst exponent is regressing to the mean...

Knightravine · Post by **Knightravine** » Fri Dec 03, 2010 2:14 am

Please tell me if this looks like an understanding of hurst.

•Persistence - value 0.5 - 1 (i.e. whatever is happening now is likely to continue)

•Anti-persistence - value 0 - 0.5 (i.e. whatever is happening now is likely to reverse)

•Randomness - value around 0.5 (i.e. likely to go in any direction)

series {1,2,3,4,5,6,7,8,8,7,6,5,4,3,2,1}

Mean is: ~4

Hurst Estimation:

input length = 8;

def price = series

def RSq = Inertia(

Sqr(correlation(barNumber(),

price, length)

), length) /

Inertia((length / 2), length);

Knightravine · Post by **Knightravine** » Fri Dec 03, 2010 3:43 am

Seems like Hurst Expontent only visually looks promising on tick charts. Is there a reason that I would abduct/intuite/guess that? Any facts?

FDAXHunter · Post by **FDAXHunter** » Fri Dec 03, 2010 8:12 am

Wow... this is what we wasted our time on?

Knightravine · Post by **Knightravine** » Fri Dec 03, 2010 9:44 pm

Yes we do. Please send me in a new direction.

FDAXHunter · Post by **FDAXHunter** » Fri Dec 03, 2010 10:54 pm

knightravine: Please send me in a new direction.

Okay: go north-east.

Knightravine · Post by **Knightravine** » Fri Dec 03, 2010 11:06 pm

I kept going to my right and I got dizzy and fell down. I am posting on this phorum because I want to know if I am starting in the wrong place with things such as Kelly criterion and Hurst exponent. I need a starting point. A few key words. I will leave your board free of my garbage. But I just want a scent of a trail. I'm curious about what you do; but not that serious about it. So, if you will, point me in a real direction. Thanks and farewell.

jslade · Post by **jslade** » Sat Dec 04, 2010 11:23 am

---> Blendertime

Go read a book on Turtle trading. Then go read a book on information theory. I think it's fair to say, most people do something in between. FWIIW, I thought some of the other dudes comments on Hurst exponent were kind of neat, even though I think Hurst is silly.

Reactor Core · Post by **Reactor Core** » Sat Dec 04, 2010 11:30 am

This thread was moved from [url=/Show%20Forum.aspx?ForumIDKey=3]Pricing & Modelling[/url] to [url=/Show%20Forum.aspx?ForumIDKey=12]Trashcan[/url].

Thread moved at request of participants.

Knightravine · Post by **Knightravine** » Sat Dec 04, 2010 10:32 pm

Kelly provided a good basis for a risk based trailing stop. If I take hurst around 0.05, it causes the odds to set to 50/50 else I default to odds I get from back testing. Simply, hurst around 0.05 makes it unlikely that any price target will be me met during the time period hurst0.05=true. (Tried H<average (H) not as good as 0.05) The edge comes from some ratio of support and resistance... get that recursively (my name for wavelet stuff) using vwap+low. The visual representation has an unexpected hyperbolic appearance.   I graph stuff because it sinks better with my mind and I can make visual judgments much faster than doing matlab or excel or t-sql or C#   The quicker I try and fail the faster I get something I like.

Isalde, this is a very succinct statement: "Turtle trading. Then go read a book on information theory. "   Shows me I'm already on a good track.

Abram Moles has the best information theory book. It is in french and english.   I tried to apply information theory and recursion to Diffusion studies (spread of ideas.) I was surprised information theory was used in Finance. I cannot believe I was blind to that.   Finance is less fuzzy than “human behavior.”

Seems like the trash is a good place to be here. Back to wasting time. So long.