how do random entries perform on the random price curves ? let us perform basic TA work on the random curves, looks like a trend to me.
trend following delusion shattered
- Thread starter hank rollins
- Start date
- Status
how do random entries perform on the random price curves ? let us perform basic TA work on the random curves, looks like a trend to me.
I asked first.Quote from hank rollins:
how do random entries perform on the random price curves ? let us perform basic TA work on the random curves, looks like a trend to me.
my initial hypothesis is that one can not tell the difference visually, by using TA, from a random trend and an apparent trend generated from actual price. am i close, mr. sub ?
I posted the 2 charts so that you would get a feel for them, and not to prove that they were "visually" indistinguishable.
I don't wish to misinterpret your statement. Are you saying that it is impossible to exploit my randomly generated "trends", as I asked in my original question?
Gentlemen:
Values generated randomly (as opposed to psuedo randomly) are going to fit both descriptions. As long as you generate a significant sample size, you will see both trending and non trending lines. In addition, you will see trending and non-trending line "segments". Depending on your ability to see the lines under higher or lower resolution, this is clear to the human eye.
I think Mr. Subliminal, that there is another element that no one, even the learned Mr. Marketsmurfer, and Nickelburper have not talked about. The fact is that trend "happens". You can see it. Even though I can describe it easily and we can see it, it can be difficult for the trader to take advantage of. The reason is "stationarity". If trend exists because everyone is jumping on board due to a favorable economic report that is one thing. If trend exists because you have generated it with a random number generator it is entirely another. The difference is that I can anticipate one, but not the other. Therefore I can trade the trend that I know exists for a reason with a little more confidence. I can stay with my position a little longer because I may know (from seeing price action during previous reports) just about how long this "trend" is likely to persist. With a randomly generated trend, I cannot know if or when price will go back to random chop. So in conclusion, I can easily test the previous day or week to determine whether in that period, price exhibited stationarity or random behavior. Once I have that answer, I can anticipate the next time period, having good reason to believe that it may continue in the same fashion. Of course one realizes that conditions can change "on a dime", but so far (over a period of about 10 years), the general principles of parametric statistics have worked out reasonably well.
Take Care
Lefty
P.S. by the way, to anwer Mr. Subliminal's question clearly. If one were to successfully "exploit" trend that is randomly generated, the result would have been obtained by "luck". That is the "A"answer.
Values generated randomly (as opposed to psuedo randomly) are going to fit both descriptions. As long as you generate a significant sample size, you will see both trending and non trending lines. In addition, you will see trending and non-trending line "segments". Depending on your ability to see the lines under higher or lower resolution, this is clear to the human eye.
I think Mr. Subliminal, that there is another element that no one, even the learned Mr. Marketsmurfer, and Nickelburper have not talked about. The fact is that trend "happens". You can see it. Even though I can describe it easily and we can see it, it can be difficult for the trader to take advantage of. The reason is "stationarity". If trend exists because everyone is jumping on board due to a favorable economic report that is one thing. If trend exists because you have generated it with a random number generator it is entirely another. The difference is that I can anticipate one, but not the other. Therefore I can trade the trend that I know exists for a reason with a little more confidence. I can stay with my position a little longer because I may know (from seeing price action during previous reports) just about how long this "trend" is likely to persist. With a randomly generated trend, I cannot know if or when price will go back to random chop. So in conclusion, I can easily test the previous day or week to determine whether in that period, price exhibited stationarity or random behavior. Once I have that answer, I can anticipate the next time period, having good reason to believe that it may continue in the same fashion. Of course one realizes that conditions can change "on a dime", but so far (over a period of about 10 years), the general principles of parametric statistics have worked out reasonably well.
Take Care
Lefty
P.S. by the way, to anwer Mr. Subliminal's question clearly. If one were to successfully "exploit" trend that is randomly generated, the result would have been obtained by "luck". That is the "A"answer.
Lefty, when I talk about random number generation on a 'puter, I naturally assume it to mean pseudo random number generation. Hence the omission of the "pseudo". Sorry if it caused confusion.
Please, no name-calling.
Ditto.
You're welcome.
That was not the intention.
Human factors aside, would you be willing to trade my randomly generated trends?
it is as impossible as exploiting any perceived trend consistently.
sometimes the trader will catch the move resulting in profits. other times the trader will miss the move resulting in loses.
- Status