direct statistical trading a "clearly" defined approach by NTW31

nukethewhales31 · Oct 17, 2009

Quote from stevegee58:

Hey nukes, have you been able to find tradable edges using tick data with the original concepts in the thread? I've written a program that implements your original LSD/SLD algo on OHLC data and it seems to work pretty well. Later I wrote another program that crunches tick data to do the same statistical analysis. Problem was it didn't have an edge, basically just a coin toss.

It seems strange that there's an edge with OHLC data but not ticks.

wierd can you post the equity curve for the ohlc data and for tick data.. tick data is more efficient generally from my use but use what works... but yeah id love to see the equity curve you got out of you OHLC algo.

nukethewhales31 · Oct 17, 2009

Quote from kiwi_trader:

Thats interesting Steve.

It would indicate to me that the bars are being watched by enough decision makers to generate said edge.

bars really are the only thing watched by most traders whether they want to admit it or not if you 7 bars in a row down into a retracement at 140%+ then someone is gonna be neverous about selling.. its psyc. even if you dont realize it happening.

nuke

nukethewhales31 · Oct 17, 2009

Quote from TheRumpledOne:

Hi Nuke:

Long time, no hear.

Are you back in the saddle?

I still have that code I was working on.

Im always in the saddle lol i just wanted to try a couple of things ... they failed miserably ... i realized i should just stick to what im good at lol.. but yeah im always here i was thinking about putting up a web page(free of course) with videos on some systems and MM and stuff I use and how to add this analysis to trading systems like pin bar/PA stuff fibs/ indicators to make them "work".. and piece things together for some people.. there is a whole lot written everywhere about setups and stuff but nothing that makes any of it work

flyingdutchmen · Oct 18, 2009

Quote from nukethewhales31:

how is it working out for you flying?

Hey Nuke good to see you back, havent seen you in a while.doing fine i hope?

the concept is clear now and i am still
working on it. there have been some things i did not unstand at first but after re-reading
a couple times the first 40 pages of the tread (ouch) i believe to have full understanding of the concept. i will
post some equity screenshots soon. i love the fact there are no parameters to overoptimize exept N, it is just all so simple and more important "logical"

thank you again

nukethewhales31 · Oct 23, 2009

Quote from flyingdutchmen:

Hey Nuke good to see you back, havent seen you in a while.doing fine i hope?

the concept is clear now and i am still
working on it. there have been some things i did not unstand at first but after re-reading
a couple times the first 40 pages of the tread (ouch) i believe to have full understanding of the concept. i will
post some equity screenshots soon. i love the fact there are no parameters to overoptimize exept N, it is just all so simple and more important "logical"

thank you again

yeah im doing good workin on some new strategies, working on a webpage and stuff, yeah the parameters shouldnt allow you to over optimize to much but it is possible gotta be careful.

flyingdutchmen · Oct 23, 2009

Quote from nukethewhales31:

thats correct same idea but this time we dont need to swap lsd's or sld's as tick data is very clear of its movements and only goes in one direction

Nuke,
you have said when using tickdata there would be no need to swap SLD's and LSD's.could you explane please

when using tickdata one would be able to determine how far price has moved from open before returning back to open, wouldnt this just add one more variant of SD and make the LSD the highest number of the usual SD and the Amount of pips/ticks that have been reached from open before returning to open again ?

lets say a bar has
open at 1,5000
high at 1,5050
low at 1,4990

and wenn using tickdata we have found that price went from open 1,5000 to 1,5020 before returing to open and making the high

so for this bar we would have a LD of 50pip ( L-O )

an SD of 10pip ( O-L ) but we have seen price going 20pip up before returing to open so this should make an SD the highest of MAX( 10, 20 ) = 20 pip, is this correct?

why is there no need anymore to swap any SD that is (and still could be) bigger than the ShortestLongestDistance over N-bars ?

where am i wrong in my thinking,
isnt it still possible to find higher SD's from last bar than N related SLD's ?

what has to be done to eliminate the 20% statistical failure

stevegee58 · Oct 24, 2009

Hey nukes, I mentioned earlier that I was having problems with finding a statistical edge with the tick data. I spent some time this morning making some improvements to my program and I'm getting some good results now.

Just needed a little massaging with the entry and tp thresholds.

stevegee58 · Oct 25, 2009

I was able to score a humongous tick data file (around 4 gigs) in csv format. How do you manage your data? You'd said earlier that you process tick files yourself.

I've been trying to import the data into a database...

Snaggleteefs · Oct 26, 2009

steve, did you get it from the eoddata website?

scJohn · Oct 31, 2009

I have a question about removing the 'extreme trades'.

I am using 80% accuracy with N = 21 so I should remove 4 trades from SD and/or LD columns that will maximize the difference between SLD (smallest largest distance) and LSD (largest smallest difference). Now let's say than in the LD column I have the shortest/smallest value of 8. However this value of 8 occurs twice. I delete both instances of 8. Now does deleting both instance of 8 count as 1 or 2 toward my removal of 4 trades? Does this effect the calculation of NPOT?

thanks.