Optimal values of SL and TP orders for an arbitrary TS. - page 3
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
Мне нужна критика, это единственный способ не попасть в дурнушку.
You mean this enumeration:
Probably each of us has at least once wondered what values of protective orders should be chosen for the reliable operation of the created Trading Strategy (TS). Some of them advise to use TP=SL and not less than 100 pips, and others advise to use TP much larger than SL - thus following the strategy "to let profits grow and cut losses". Others prefer scalping with short targets (TP<100 pips). So which trading strategy to follow?
There is nothing to criticize. None of your methods is self-sufficient, what should be criticized separately from the used TP. It's obvious. TP and SL are two sides of one coin, for each input one maximizes the income (TP output), the other minimizes the cost (SL output).
You see, there is nothing to criticise. Your question is like the great Thinker "42"'s answer to the question "What is the meaning of life and all". If you want criticism - give at least "one unknown" :o)
PS: For each specific TP a specific SL is assigned and no other way around, this is packed into an order and lives together for a while. But maybe we should think in the opposite direction - first we define SL and set TP for it. Anything is possible.
I suggest the following concept of "communication" with the market.
Strangely enough, this block for some reason I did not notice at once. Seryoga, maybe you're completing it slowly :o). I'll have to think about it.
So, perhaps we should criticize here:
Предлагаю следующую концепцию "общения" с Рынком: Котир->ТС->ММ->$ Суть сводится к тому, что мы создаём набор правил по которым ТС оптимальным образом нарезает ценовой ряд, максимизируя доходность, которую я определяю как количество пунктов, которые зарабатывает ТС за единицу "человеческого времени". Далее, блок ММ максимизирует процесс перевода этих рыночных пунктов в рублики (оптимальный ММ заточенный под конкретную ТС). Всё.
Let's take as a base the fact that no MM will be able to gain positive output profit in rubles if the expected payoff (ME) of the TS is less than zero (we will take as ME the average number of points coming to one transaction, taking into account brokerage companies' commission). This is a very complicated task, which is much more difficult than searching for an optimal MM, and it seems to me that nobody has paid much attention to this very task - they focus on the searching of optimal parameters of an arbitrary TS. It is obvious that the number of arbitrary (those that can be invented) TS is infinite, and their parameters are even more (if, of course, one can put it that way) and the problem does not converge in principle, i.e. no life is enough to try all strategies that can come to inquisitive mind, and certainly to try all their tuning parameters in the tester. Therefore it seems we need a systematic approach to choosing the optimal TS and its best operation.
Also, let's say right away that
1. no TS is able to get MM different from zero at martingale (integrated random variable (IC) with zero MM - at first approximation, an analogue of price series) with a rather long history.
The latter reservation is important, because any results are possible on short historical samples, which are by their nature statistical fluctuations and have nothing to do with reality. For this reason every now and then there are periodic rumors in trading circles about "miraculous" enrichment of lucky friends. All these rumors are based on the fact that in life, as a rule, people often exaggerate successes and keep silent failures. This gives the impression that everyone around you is busy making money and that you are a loser.
2. Miracles do not happen (in all their forms).
Nevertheless, I begin my consideration of these problems with optimal MM for arbitrary TS, which is given with a sufficiently large set of tricks (or, similarly, the first difference (FDR) of the account balance). It's easier for me to present the material that way.
Now, Seryoga, poke your finger where you're proposing approaches to optimal SL and TP values here? Vt is this:
Kotier->TCs->MM->$.
Or maybe here:
which the TS slices the price series optimally, maximising the return, which I define as the number of pips the TS earns per unit of "human time". Then the MM block maximizes the process of translating these market pips into rubles (optimal MM sharpened for a particular TS). That's all.
Seryoga, I understand that you opened something new? The "MM will maximize the process of transformation of these market points in rubles" - is it necessary to criticize? In short, have fun on your own, I will not disturb you :o)
Yes... reading the thread and realizing that if there are still romantics in forex, they must be only neuron programmers. In this case, the market is not the only one, but the market is the only one, and the market is not the only one, and the market is not the only one.
The topic itself consists of two questions: determining the optimal price level SL (TP) and determining the optimal amount of funds for this trade.
For some reason no one has said a word about the theory of probability. If SL<TP then for a stationary process the percentage of losing trades will be less than profitable ones, but the impact on the account of an average losing trade will be less than the impact of an average profitable trade. Thus, the resulting influence of profitable and losing trades will be neutral. The same applies to the TP<SL ratio. In this case the profitability of the system will be more than 50% even on a stationary process. I.e., it's absolutely all the same for martingale with the ratio of SL to TP; the resulting impact will always be the same and equal to 50%. Since you develop a universal model of SL and TP setting, you should rely on the universal market model, and the universal generalized and simplified market model is the martingale. And if this is so, then it means that you contradict yourself, because for martingale there is no difference in terms of risk levels and profits, as well as the volume of transactions, which automatically means that the creation of a universal model is impossible in principle.
The rational working solution is to find the optimal SL TP for a particular TS and for a particular market. Here's a specific example: the robot enters the market at the opening of the day, sets SL equal to the value of ATR and exits at the opening of the next day:
Something like a trend appears.
Now let's change the condition. The robot has held the position for five days:
Well, it seems to have let profit accumulate, but the result is much worse!
Here's another TS tested in the same market:
The robot held a position for one day, and the result was not very good. We return to the concept "cut losses, increase profits", namely, we keep the position for five days:
As we can see, profits are increasing. The quality of modelling does not change the results of these strategies and in the second strategy stops were not used at all.
As we can see, what is fun for one TS, is death for the other.
in general, I think, in a "favorable" situation, SL should be (can? preferably? possibly?) taken not as a hard order to close, but as an order to find the minimum losses, upon reaching certain levels calculated from the Calculated SL... again, we need specifics... ( considering this issue leads not to a formula for calculating SL and TP, but to new strategies! which is undoubtedly a more positive effect...)
А теперь, Серега, ткни пальцем, где ты тут предлагаешь подходы к оптимальным значениям SL и TP? Вт это что ли:
или может быть тут:
Серега, так понимаю, ты открыл что то новое? "ММ максимизирует процесс перевода этих рыночных пунктов в рублики" - это нужно критиковать? Короче, развлекайся самостоятельно, не буду тебе мешать :о)
Sergei, you should... ...have a beer or something - chill out a bit, I haven't said anything yet. It's all a fairy tale - we're in no hurry, are we? Be patient, I'll prepare material, systematize it, post it and then I'll give you a start on your critics :-)
C-4 wrote(a) >> For some reason no one has said a word about probability theory. If SL<TP then for a stationary process the percentage of losing trades will be less than profitable, but the impact on the average losing trade will be less than the impact of the average profitable trade. Thus, the resulting influence of profitable and losing trades will be neutral. The same applies to the TP<SL ratio. In this case the profitability of the system will be more than 50% even on a stationary process. I.e., it's absolutely all the same for martingale with the ratio of SL to TP; the resulting impact will always be the same and equal to 50%. Since you develop a universal model of SL and TP setting, you should rely on the universal market model, and the universal generalized and simplified market model is the martingale. And if this is so, then it means that you contradict yourself, because for martingale there is no difference in terms of risk levels and profits and transaction volume, which automatically means that the creation of a universal model is impossible in principle.
Everything is correct, except one - I do not consider the process of quoting as martingales when I speak about the possibility to earn statistically in the market (TS with MO>0) and I consider it as such when I consider approximations that allow obtaining analytical expressions for parameterization of TS algorithm's operation. At the same time I carefully estimate possible errors related to the use of one or another assumption. Therefore, there is no contradiction, unless of course I inadvertently make a mistake. This is what criticism is for.
Серёга, ты енто... пивка что ль глотни - охлонись немного, я ещё ничего не сказал. Это всё присказка - мы же с тобой никуда не спешим? Потерпи, я материал подготовлю, систематизирую, выставлю и тогда дам тебе отмашку на критику:-)
How you didn't say anything, then who have I been talking to all this time? :о) All right, but you tell me... tell me when it's okay to criticize, because I'm losing my Belinsky.)
Как ты ничего не сказал, а с кем же я тогда беседовал все это время? :о)
Well, it reminded me of shadow-fighting too... In short, don't be so hard on yourself - I need it too :-)
ivandurak wrote(a) >> Maybe it's more productive to ask another question . How long SL and TP will be relevant, at least a qualitative estimate, as well as the criterion for a new search for TC, optimization, etc.
Don't get ahead of yourself. We should talk about dynamic parameters looking at the analytical expressions into which they are included by parameters. It is more clear and less subjective and, therefore, closer to reality.
It may be more productive to ask another question . How long the found SL and TP will be relevant, at least a qualitative assessment, as well as the criterion for starting a new search for TS, optimization, etc.
From my point of view, I would answer this question by saying that "stops live as long as the trade is calculated". i.e. that means different stops for each trade. it's another matter if they "happen" to coincide with the stops of previous trades - that's another matter.
PS: ............ Но может имеет смысл думать в "обратном" направлении, - сначала определяться с SL и для него выставлять TP. Все возможно.
First determine the SL, calculate the volume from the chosen SL, then only think about the TR. In my opinion, this is the only way to do it.