Physics Help Forum Hurst exponent in Tick Volume.

 Physics Physics Forum - General Physics Discussion and Physics News

Nov 13th 2018, 01:43 PM   #1
Junior Member

Join Date: Nov 2018
Posts: 2
Hurst exponent in Tick Volume.

Hello,

I am trying to calculate Hurst exponent using as an input FX Tick Volume. However being not educated in advanced math and physics, I struggle to obtain clear and objective results. I am not sure whether I get reliable Hurst exponent calculation.

One Tick is each change of price, which can occur a few times a second. As there is no source for the Real Volume(in USD etc.), Tick Volume is the only reliable data available for research purpose.

What I do is I take a sample of Tick Volume(below you will find just a fraction, I attached full sample in file "DATA"):

1.14128
1.14133
1.14127
1.14132
1.14138
1.14142
1.14136
1.14141
1.14137

and then I change it into a binary system, tick higher(price higher) than previous=1 and lower(price lower) than previous=0.

I get:

1.14128 1
1.14133 1
1.14127 0
1.14132 1
1.14138 1
1.14142 1
1.14136 0
1.14141 1
1.14137 0

Finally, I put "0/1" results into excel sheet with Hurst formula(please see attached file "HURST FORMULA", which I found on the internet).

My questions are:

1. Those it has any sense to change prices into a binary system?

2. Those my Hurst formula provides/could provide objective results, should I use a different method?

3. Maybe there is any software to calculate Hurst Exponent?

4. Maybe there something I do not know concerning Hurst Exponent calculation, and one could enlighten me?

Attached Files
 HURST FORMULA.xls (753.0 KB, 1 views) DATA.xls (59.0 KB, 0 views)

 Nov 14th 2018, 06:01 AM #2 Senior Member     Join Date: Jun 2016 Location: England Posts: 929 I am not familiar with the Hurst Coefficient, but I had a quick look on Wikipedia. It is a mathematical method for determining the self correlation of a sequence of values. 1.0 to 0.5 indicates that increasing values tend to be followed by increasing values 0.5 to 0.0 indicates that increasing values tend to be followed by decreasing values (=0.5 indicates random numbers). I would not say that it requires advanced mathematics (perhaps intermediate level) but I would suggest that some study of the (statistical) considerations behind it would be useful. Currently you seem to be just pushing numbers in without really understanding, and expecting the output to be meaningful. topsquark likes this. __________________ ~\o/~
Nov 14th 2018, 08:34 AM   #3
Junior Member

Join Date: Nov 2018
Posts: 2
 Originally Posted by Woody I am not familiar with the Hurst Coefficient, but I had a quick look on Wikipedia. It is a mathematical method for determining the self correlation of a sequence of values. 1.0 to 0.5 indicates that increasing values tend to be followed by increasing values 0.5 to 0.0 indicates that increasing values tend to be followed by decreasing values (=0.5 indicates random numbers). I would not say that it requires advanced mathematics (perhaps intermediate level) but I would suggest that some study of the (statistical) considerations behind it would be useful. Currently you seem to be just pushing numbers in without really understanding, and expecting the output to be meaningful.