Optimal values of SL and TP orders for an arbitrary TS. - page 8
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
>> Neutron
Dynamic TS has long been my practice rather than a hypothesis ... the lifetime of a movement is not as important as its frequency of occurrence and the uncertainty with which it can be determined ... such "piecemeal" systems also have their place ... of course an Expert Advisor on one codec can sit and wait for weeks to work out 1 position... ... of course it's a very time consuming task, a dozen codecs need to be found and working in parallel, each one taking its own piece ...
>> C-4
What do you mean by grail? ... The system which eventually gives a permanent bonus in a month is the Grail? Profit is floating of course ... for example 2-10% - is it a grail ? ... if yes - there are a lot of such Grails ...
>> Neutron
If you don't want to trade robots you'll have to raise some profit and defeat them ... If you don't want to trade robots you'll have to raise some profit and defeat them yourself ... If you don't want to trade robots you'll have to raise some profit ...
Getting ahead of ourselves again)
RIV писал(а) >>
>> Neutron
... of course it's a very time-consuming task, you need a dozen of codecs to work in parallel and each take its own piece ...
And how do you "dig"?
Сергей, однако такое разбиение на три области не учитывает один существенный эффект. Часть прибыльных ордеров прежде чем закрыться успевает побывать в зоне SL (аналогично и с ТР). То есть введение SL (как и TP) как правило существенно деформирует всё распределение, включая зону штатных выходов ТС. Или я забегаю вперёд и дальше ты опишешь и этот эффект?
Hi Nikolai!
Very subtle point on your part. Indeed on the real TC bribe distribution you can observe finite areas of "gaps" in the form of deformation of lugs. This manifests itself in their widening and reduction in height compared to the model problem. I evaluated the error introduced by the limiting transition to narrow "lugs". It turns out that the problem includes the area occupied by an ear, and it miraculously does not depend on its width (the wider it is, the lower it is). And in the first approximation error does not accumulate.
Thanks for the valuable remark.
Farnsworth wrote >>
Again, just to be sure - what you have drawn, the distribution and the MO taken from the ceiling is derived by which option:
The points that only TC brought in
points brought by TC and joint SL/TP triggering
Just, a lot of questions, for example - how did you get the distribution only TC? Did you get it with infinite deposit on infinite history?
Only now, Sergei, I understand what bothers you - you are approaching the understanding from a practical point of view (like, how to use it all in practice, if there is only the data from the trading account), and I try to abstract the maximum, to get even though "ideal" solution, but still a solution to the problem.
So. I have a "clean" TS without stoppers and spreads working with a real symbol (open prices, minutes). It gives me its bribes in pips without slippages and any other real stuff (spherical horse in a vacuum). And only then I put into play all the necessary features that bring the analytical model as close to reality as possible. I believe this approach can lead to effective results.
I would like to note that the effect of protective stops on FR shown in the picture above is universal and does not depend on the particular type of TS itself. Therefore, it is enough to mark out the middle part of FR (it doesn't react to SL & TP orders in the system) to restore the whole FR unambiguously and plot three logarithm profit integrals for further analysis.
And another thing, Seryoga, you must have tested your numerous systems based on Pastukhov's dissertation on a large volume. Can we have a look at their real distribution? I think there should be enough orders there (but of course not a fact).
You can. A little later.
A dynamic TS for me has long been a practice rather than a hypothesis... it is not so important the lifetime of a movement character as its frequency of occurrence and with what error it can be determined... such "piecewise" systems also have their place
...Something about the short lifetime of detected patterns does not work for me. The point is that the time required to identify them is, on average, equal to their characteristic lifetimes. But I agree that a more detailed research approach is needed. State your concept, RIV.
Действительно на реальном распределении взяток ТС наблюдаются конечные области "залётов" в виде деформации ушек. Проявляется это в их уширении и уменьшении высоты по сравнению с модельной задачей. Я оценивал ошибку которую вносит предельный переход к узким "ушкам" . Получается, что в задачу входит площадь занимаемая ушком, а она чудесным образом не зависит от его ширины (чем шире - тем ниже). И в первом приближении ошибка не накапливатся.
There is a curious effect with the lug area. It turns out that the shift in the MO due to a change in the shape of the distribution is well approximated by this simple approximation.
I have a bridge to trading contexts suddenly starting to emerge. With typical value of H-volatility "almost" 2 and TP > SL, introduction of SL for +SL bribe size will cut off (transfer to the left ear) about a half of events, for bribes larger than +SL the loss will be even larger, for bribes in the -SL, +SL interval transfers to the left and right ears will partly compensate each other. The result will generally be some, sometimes very strong, displacement of the MO.
Now suppose that TC is oriented to contexts with H significantly different from 2, and is somehow able to guess them. Obviously, the processes of events transfer to ears (and hence MO drift) will be substantially suppressed (with respect to the case H = 2).
>> storm >> Neutron
You guys are modest... :) ... at least you didn't ask for money ... :)
I think everyone must contribute his part of the effort and expense ... time, money, health ... to get a result ... as in any other business ... and not to produce haplessness ...
>> Neutron
Naturally everything is determined with a margin of error ... otherwise it would be just an unreal freebie ...
Now suppose that the TM is oriented to contexts with H significantly different from 2, and is somehow able to guess them. Obviously, the event transfer processes in the ears (and hence the MO drift) would be significantly suppressed (relative to the case of H = 2).
At the time, unfortunately, of the former activity of Sergei's previous thread, I investigated the cagey patterns of the n-long one. It turned out some interesting "side" conclusions, e.g:
- there are "converging to" and "diverging from" 2H patterns with considerable support and interest,
- And most interestingly, imho, they have a large "binding" to "external" marks, e.g. time ( which is understandable, for example, for relatively short-lived patterns)
From here we can try to make a connection with the ratio of SL and TP
At typical H-volatility "almost" 2 and TP > SL, SL on the +SL bribe size will subtract (transfer to the left ear) about a half of events; for bribes larger than +SL the subtractions will be even larger, and transfers to the left and right ears will partly compensate each other for bribes in the -SL, +SL interval. The result will generally be some, sometimes very strong, shift in the MO.
Can you try to "formalise" this assumption?
В пору, к сожалению, былой активности пердыдущей темы Сергея поисследовал каги-паттерны n-длинной. Получилось несколько интересных "побочных" выводов, например:
- есть "сходящиеся к" и "расходящиеся от" 2Н паттерны со значительной поддержкой и интересностью,
- и самое интересное, имхо, у них большая "привязка" к "внешним" параметрам, например времени ( что и понятно, например, для относительно непродолжительных паттерн)
Отсюда можно попробовать сделат связку с соотношением SL и TP
Can you try to "formalise" this assumption?
No, I can't do that now, it's not the right form.
P.S. Yeah, I think I found a discussion, thanks to your few posts)
to Neutron
Только сейчас, Серёга, я въехал что тебя настораживает - ты подходишь к пониманию с практической стороны (типа, а как это всё на практике использовать, если имеются только данные с торгового счёта), а я пытаюсь абстагироваться максимально, для получения пусть и "идеального" решения, но всё же решения поставленной задачи.
I try to be both a theorist and a practitioner. I recommend it :o))
So. I have a "clean" TS without stopper and spread working on a real instrument (opening price, minutes). It gives me its bribes in pips without slippages and any other real stuff (spherical horse in a vacuum). And only then I put into play all the necessary features that bring the analytical model as close to reality as possible. I believe that this approach can lead to effective results.
So you're taking a theoretical distribution of the TS trade? And how do negative trades turn out in theory? It turns out that TS closes the deal at a loss by itself? And what is the time of waiting or by what SL it closes? Here you should not pay due attention to my comment: "How did you get the distribution of the TS only? What have you got it from, an infinite deposit on an infinite history?" That's where my misunderstanding lies. I thought you wrote that the TS defines entry and exit. Was it meant and exit by loss? What is it?
PS: I'm just not familiar with your TS and I'm asking probably stupid questions.
I would like to point out that the effect of protective stops on FR presented in the picture above is universal and does not depend on the specific type of the TS itself. Therefore, it is enough to mark out the middle part of the FR (it does not react at all to SL & TP orders in the system) to restore the whole FR unambiguously and plot three logarithm profit integrals for further analysis.
And why don't you "cut" from (-) to zero in the picture above, making the strategy super profitable? And do answer - does everything you've written refer to protective orders or to trade order parameters. It seems to be related to the parameters, but sometimes you use the word "protective order".
to Candid, Neutron
Sergei, however, such a breakdown into three areas does not take into account one important effect. A part of a profitable order has time to be in the SL zone before it closes (the same thing with TP). It means that introduction of SL (as well as TP) deforms considerably the entire distribution, including the zone of standard TP exits. Or I am getting ahead of myself and further you will describe this effect?
So, apparently I'm again the only one who doesn't understand all the subtleties :o) Where does this effect appear in the TC distribution? It seems to fix the fact - (+) or (-) when trading, without any regard to the drawdown. How the fact that a profitable trade was losing for some time and vice versa will affect the distribution form (by fixing points)? It's not even anywhere in the reasoning at all.
In general I doubt the efficiency of astrolabe so far, and if we include f variable and drawdown calculation and really decide about FR, I am afraid the joy will be less.
Although, of course, I will not be in a hurry.
Каким образом тот факт, что прибыльная сделка некоторое время была убыточной и наоборот скажется на форме распределения (по факту фиксации пунктов)?
If there is no stop, it will close in the positive and fall into one bar of the histogram. If we enter a stop and the price catches it, it will close in the negative and hit another bar in the histogram. In the second case, it will go to the leftmost one (ear).
By the way, happy birthday to you :). Although they already happened yesterday :)
Если нет стопа, она закроется в плюс и попадёт в один столбик гистограммы. Если мы вводим стоп и цена его зацепит, она закроется в минус и попадёт в другой столбик гистограммы. Во втором случае конкретно в крайний левый (в ухо).
That is clear, but what is not clear is your statement:
A part of profitable orders has time to be in the SL zone before they are closed (the same is with TP). That is, the SL (as well as TP) will deform significantly the entire distribution, including the zone of standard TP exits
How does this "been there" or "been there" deform the graph. You get a fact and that's it. Besides, we have two distributions (mind you):
Which one was it? Sergey wrote about the first one, but there are no stops there. But it turns out that we have two levels of stops? As you understand, I'm a little confused. And it's also not clear what's up with the "protective orders", what are those things? And don't hurry with the deformation - let's look at the real distribution - there you will show me what has become bent and where the ears have grown.
By the way, happy birthday to you :). Although it already happened yesterday :)
Thank you! Real science is done by 160 - year old senile people! http://www.futurama.ru/farnsworth.shtml :o)