Why do I see "Trends" in Randomly Generated Data?

in the end what matters is that you design a simulation of trading technique via programming the technique then....
running the simulation on one market for 4 years, and if it is statistically tradeable, then apply the final acid test, run it on a completely different instrument over a different time period, but at least 4 years. If the second set produces stats within 70% of the first set you have a viable EDGE, and the road to nirvana and riches is yours. But can you find such a system?
and remember tradeable stats means reasonable drawdowns against net profits, i.e., a ratio of 10/1 or more.
So find me such a system if you can.
 
I have not read the whole thread so maybe someone already mentioned that there is no such a phenomena as a pure random number generator and all of them follow the thought process and the mind pattern of their creators which is why they naturally exhibit a pattern.

q.
 
Quote from ML_QUANT:

I have not read the whole thread so maybe someone already mentioned that there is no such a phenomena as a pure random number generator and all of them follow the thought process and the mind pattern of their creators which is why they naturally exhibit a pattern.

q.
More...

No, there are so called hardware random number generators. Software (program code)? Yes, it is deterministic, but since most of the algorithms use the so called "seed number" and most of the time it is taken from computer's clock (nanotime), they are not as non-random as you might imagine. More advanced software random number generators use more sophisticated methods for generation. For instance, some RSA key generators take for input such things as HDD, keyboard and other hardware variables to make generation as random as possible.

Hardware generators are even more reliable. They use for input noise from electric circuits, radiation source from nuclear decay and many other phenomena you haven't thought of.

Seriously, this "computer does not generate random numbers" argument is so trivial that it borders with such arguments like "what if my CPU messes up one single bit somewhere sometime? therefore I don't trust CPUs, and please ban usage of computers for medical and nuclear research".

What you're talking about is true. That's why term "pseudo-random" is used sometimes. But as I've already stated this problem is quite trivial. If it wasn't then you wouldn't be able to trust cryptography, monte carlo simulations and much much more stuff that is used daily in almost any industry worldwide.
 
Quote from ML_QUANT:

I have not read the whole thread so maybe someone already mentioned that there is no such a phenomena as a pure random number generator and all of them follow the thought process and the mind pattern of their creators which is why they naturally exhibit a pattern.

q.
More...


Of course, but some pseudo-random numbers are, for some purposes, random enough to work.
 
I would like someone to explain to me what the purpose of the time and effort is, that is attributed to creating and watching randomly generated data in fictious environments.
 
Quote from ProfLogic:

I would like someone to explain to me what the purpose of the time and effort is, that is attributed to creating and watching randomly generated data in fictious environments.
More...

It's called the "academical approach". Some people get stuck in any attempt to get ready to test a system. The only problem is that they only discuss about the definition of words and the value of the data to be used to test. That's the farest they get. Their purpose is not to find or test a system; they only focus on finding the ultimate definition of the word "random" and on the fact whether random is random or not.

There is also the "practical approach". You take a few years of real data and do a test of a few thousand trades. You analyze the results statistically and you will know if the system is valid or not. Instead of breaking your head on the problem of random or not random data you break your head on the application of the system on real data.

The practical approach is used to trade succesfully.

The academical approach is used to train your brain or to kill time, i see no other value in this approach. This approach is a waste of time to me as i see not additional advantage to the practical approach.
 
Quote from ProfLogic:

I would like someone to explain to me what the purpose of the time and effort is, that is attributed to creating and watching randomly generated data in fictious environments.
More...

Well, the bases for 100% of every method used to back test, forward test and predict the markets are founded on Pattern Recognition! Including 100% of any indicator anyone has ever used...hence the endeavor to recognize how patterns work, calls for understanding what other phenomena including nature, random numbers etc. do generate patterns.
For one if a true Random Generator ever existed and was exhibiting patterns, then it wold prove that we are all wasting our times here.

q.
 
Quote from ML_QUANT:

Well, the bases for 100% of every method used to back test, forward test and predict the markets are founded on Pattern Recognition! Including 100% of any indicator anyone has ever used...hence the endeavor to recognize how patterns work, calls for understanding what other phenomena including nature, random numbers etc. do generate patterns.
For one if a true Random Generator ever existed and was exhibiting patterns, then it wold prove that we are all wasting our times here.

q.
More...

Really . . . I tested my system over 7 years of real-time data and live screen time because I knew the inconsistency in the value of backtesting. It was a true pain in the a$$ testing what I do in real-time but the results were accurate and unwavering.

I guess cutting corners has a value to some.
 
Quote from spike500:

The academical approach is used to train your brain or to kill time, i see no other value in this approach. This approach is a waste of time to me as i see not additional advantage to the practical approach.
More...

I agree but just wanted to make sure I wasn't too far in left field on this and wasn't missing something new.

Thanks.
 
Quote from ML_QUANT:

Well, the bases for 100% of every method used to back test, forward test and predict the markets are founded on Pattern Recognition! Including 100% of any indicator anyone has ever used...hence the endeavor to recognize how patterns work, calls for understanding what other phenomena including nature, random numbers etc. do generate patterns.
For one if a true Random Generator ever existed and was exhibiting patterns, then it wold prove that we are all wasting our times here.

q.
More...


well stated.

s
 
You must log in or register to reply here.
Back
Top