Welcome to NexusFi: the best trading community on the planet, with over 150,000 members Sign Up Now for Free
Genuine reviews from real traders, not fake reviews from stealth vendors
Quality education from leading professional traders
We are a friendly, helpful, and positive community
We do not tolerate rude behavior, trolling, or vendors advertising in posts
We are here to help, just let us know what you need
You'll need to register in order to view the content of the threads and start contributing to our community. It's free for basic access, or support us by becoming an Elite Member -- see if you qualify for a discount below.
-- Big Mike, Site Administrator
(If you already have an account, login at the top of the page)
I'm relatively new to the concept of a Random Walk, and my first instinct was to try to generate a bunch of random data (coin toss for each tick?) and then create a dummy symbol with say, 1 year of tick data, and then backtest and optimize against that symbol in the same way I would a real instrument.
First, I am wondering if anyone has already put together the tools to create Random Walk data. Presumably it is very easy, if I understand correctly it's just a 50-50 coin toss. For me I want tick or at the very least minute data for a 1 year period. I was thinking of just doing it in C#. Has anyone done this already?
Second, I'm still trying to think ahead to the probable outcomes of backtesting and optimizing against the data, and what the benefit is. If you are of the mindset that the Markets are not random, and yet you know the Random Walk data is, then what can be learned from the process?
We know from past studies that chart experts cannot tell the difference between Random Walk data and real charts, and that both contain double tops/bottoms, head and shoulders, and all that. My original thought was to see if my strategy is curve fitted to a real instrument, whereby it would not work on the random walk data. A second thought was for a strategy to work on both the random and real data, thereby confirming (?) the strategy has some predictive capabilities.
My little contribution and I have no idea it this would be helpful for the subject...
I removed the dust on this Excel chart generator i've played with some times ago.
It use random coin toss to generate 40 000 values ....
=SI(ALEA()>0.5;1;0)) …
I'm still amazed when I look at this.....
The way I understand it ...the logic is something like:
heads: +
Tails: -
1 toss : 50% chance head : 1 point
2 toss : 25% chance to have 2 head in a row: 2 point
3 toss : 12.5% chance to have 3 head in in a row : 4 point
4 toss : 6.25% chance to have 4 head in in a row : 8 point
same thing for tails but score are -1,-2,-4,-8,-16,.....
then you had pointage
if you scored a 16 pointer (5 head in a row, you have 16pts) than you hit a tail,you start the logic back to the beginning. what's the chance you hit a -16 pointer with the tails to neutralize the +16 pointer you got ? ... poor
and you continue .... a realize that after few 1000 toss, you have a trend .... and many other figures
I'm still figuring out how to generate all the necessary data. I really prefer tick data, and 1 year of tick data is quite a bit of data, over 1GB of data probably.
I'm leaning towards just doing C# because I know it better than I know Excel.
Then I want to tie in the correlation discussion we've been having. What I envision is trying to generate two data subsets of random walk data, and then running Excel or whatever to produce a correlation factor between them.
I'll talk more about the correlation in that thread.