Avalanche - page 415

 
Diamant:

PS. in addition, pay attention to points a, b, c )) They will really get in the way of real work.

PPS. What is your testing period?

С 15.07.2010.
 
SergNF:

number of returns ... In short, I don't know what he meant or how he got such data, but this "something" is very similar to the breakdown frequency of a certain (again, which channel and how the breakdown was calculated is unknown) channel after a certain number of knees.

I wish I could estimate the uniformity of these statistics :-). And as for the method - I think such a thing should be rotated starting from every bar. And it means that in order to use it... to make channels from each bar during trading as well.
 
SergNF:


In that amount of shitting, it's not surprising.

In one of the posts from one of the Nordic countries there was a similar "tableau"

the number of returns ... I don't know what he meant and how he got such data, but this "something" is very similar to the breakdown frequency of some channel (again, what channel and how the breakdown was counted is unknown) after a certain number of knees.

I see. It seems, in short, like something out of the question.
 
jartmailru:
I would also like to evaluate the uniformity of these statistics :-). And as for the methodology - in my opinion, such a thing has to be rotated, starting from every bar. And it means that in order to use it... ...one should draw channels from every bar in order to use it.


I agree.

Therefore, in terms of forum topics, we should write a script that sets several channels on each bar and "forward" calculates similar statistics - which channel and through how many knees will be broken. At the same time certain "market parameters" are collected (on the initial bar). Except that such statistics will be "pseudoscientific". Although, probably, less "false" than statistics, which gave birth to alligators etc.

This is roughly how the coryphaei of "baroque TA" also "invented" the HeadShoulders.

(This is sort of sarcasm to the purity (or is it frequency?) fighters of the faith and a parody of some of the pathos threads/posts)

ZS.

Need to do channels from each bar when trading as well.

Or intuitively, like khorosh AND, set the width of the channel, including already when the tsepochka works, as a function ... ahem ... "market parameters". I.e. to predict changes in vAloNTilDIty ;).

In the context of the post to which I responded - simply by predicting the channel and knowing the deposit size, to determine the breakpoint of the "tripling" function, so that the deposit remains intact. True, at that "doubling in two days" won't work. that's why I don't get into "quantitative arguments" AKA bickering.

 
khorosh:
С 15.07.2010.
And in two years shall we say run?
 
SergNF:


I agree.

Therefore, in terms of forum topics, one should write a script that sets several channels on each bar and "forward" calculates similar statistics - which channel and through how many knees will be broken.

From the "agreement" :-) it follows that from the moving point, you need to set back a window width = observation period. And on this window collect statistics, which should be shown later.

Although the first phase can be as you've written - there is a starting point - the number of reversals of the channel of a given width is compared to it (it can even be with an "error" - so the 5-10 points of the 4-digit number, which are missed, are anyway considered as a touch), and then you can move by the window and count the number of reversals.

And this statistics can be used, for example, in an indicator...

And what is "pseudoscientific" about it?

 
Diamant:
How about a 2-year run?
I've posted results for 2 years in this thread before. But now I usually test on shorter intervals. For myself, I think the more important criterion of an Expert Advisor's success is how much profit it is able to draw between larger drawdowns. The 2-year test does not guarantee the possible drawdown when working on a real account. And for the EA not to lose 2 years we have to use parameters at which the EA earns much less per unit of calendar time.
 
jartmailru:

It follows from the "agreement" :-) that a window width = observation period should be drawn back from the sliding point. And on this window to collect statistics, which you can show later.

Although the first phase can be as you've written - there is a starting point - the number of turns of the channel of a given width is compared to it (it can even be with an "error" - so the 5-10 points of the 4-digit number, which are missed, are anyway considered as a touch), and then you can move by the window and count the number of turns.

And this statistics can be used, for example, in an indicator...

And what will be "pseudoscientific" about it?


The first and most important caveat :)

it follows that The question is:

To whom and why?

Next

points to back the window width = observation period

In fact there are so many parameters here that when simplified you can spout off a child too.

Before "that" statistic there was correspondence with roughly these kinds of suggestions for thought.

You have a depot, you have a favourite multiplier. Your depot will hold n knees if on each successive knee you increase the previous lot by m-fold. "Ahead" (still more illustrative than backwards) you have a channel, x-pips wide, which you will cross no more than n times.

The result is the "upper limit", i.e. if the coefficient is m, then the channel is no narrower than x. The lower bound would be (if there is) m=f(satisfying your rate of return). Then m can be made as a "function" of (conventionally) ATR or the number of knees (and see if it falls within the allowed window), you can, by changing the channel width, raise or lower the allowed m.

After all "this" (all my posts in the Avalanche/Diablo threads) is just "mental exercise" and not a discussion of TC or even TT (tactics).

And more relevant (posts) to the forum topics than this, or this, or eventhis. All the same, "a forum on mechanical...".

And what would be "pseudoscientific"?

Read the Mathematician's article about the sandwich and the non-returned Vernulllllllly, which has a coccyx (tail) larger than "3 sigmas" and is constantly oscillating (not stationary), and as a result the average temperature at xForex is 36.6(2)

(This is sarcasm about the applicability of his PhD methods to this or that material)

 
There's always a good way not to go wrong - do nothing ;-).
 
PPC:

You're going the right way, comrades. I was up to 3. With the initial deposit of 10000 (tester) the starting lot was 1.5. And so there was nothing, it held. I even got quite good profit with reinvesting in 2010 (~1400%), for 2009 with reinvesting I got ~200%, and 2009 without it I got a bit negative (my chart had some nice bumps), but I lost too much. I was just having fun with it. I have to use the handwheel to work with the fx I've already used.

I am working with the fractals for the first order by hand.