What is everyone looking for? - page 31
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
......
2) Дело вкуса - оценивай как хошь. Другой вопрос что предлагается создание некоего "универсального" теста. Для грубых оценок КПД системы.
По условию задачи тест должон показывать отношение реал/идеал. Отсюда и вылез теоретический идеал ряда транзакций в виде вершин ЗигЗага.
Yes. CD should be viewed as a flow of energy and counted as a dimensionless value of 100% in pips. Any TS can be evaluated by the number of accumulated pips, i.e. efficiency.
One and the same TS with different parameters will have different efficiency, and of course it is worth choosing parameters with higher efficiency. However, such an indicator has nothing to do with profitability of the considered TS and this is another aspect of research - it is a question of MM.
Of course we are talking about reverse TSs with consecutively going contrary to the direction of transactions.
But this is only a tiny tip of the visible and usually discussed energy of CD due to the discreteness of the TS decisions. In fact it is many orders of magnitude greater, and can be estimated by an incredibly large, but still finite number of multidirectional simultaneously open positions. And there is only one limitation - the limitation on the maximum number of simultaneously opened positions, imposed by the DC. Such a flow of numerous simultaneously opened positions would give a smooth, almost analogous "overflow" of RC's energy.
Here is the problem. In ideal TS we have 1000 trades, while in real trade we have 1400. How do you compare series of transactions with different numbers? Not to mention the fact that the inputs/outputs of an ideal TS do not correspond to those of the STS.
This is exactly not the problem. It is not different TSs that should be compared, but different parameters of ONE TS relative to an ideal ..... (there is no such concept, but the word "profitability" does not fit here, as I wrote above) of CD energy.
Programmer, it was probably your idea to compare, but why ?
Вот тут вся и проблема. В идеальной ТС - 1000 сделок, на реале - 1400. Как сравнивать ряды транзакций с разными их количествами? Я уже не говорю о том, что входы/выходы идеальной ТС никак не соответствуют оным для иТС.
Now I'd better send you to read my previous posts on this subject. Ok, I will not. I will repeat the point right here.
I suggested to write in the indicator buffer NOT trade signals, but recommended (TS) market position. I.e. the signals to be compared become rectangular, stepped or curved (if the TS trades with small gradation of lots) rather than impulse signals.
For example, while a market position is buy-1-lot - the indicator draws the line ==+1, then it closes buy == 0, opens sell-2.3 lots == - 2.3, etc.
In this variant of signaling, Pearson correlation is suitable for measuring efficiency. The second option (the obvious one) is a tester. But it requires two runs and subsequent calculations.
And here at the output - at once the efficiency according to the formula kPI = abs(correlation(ideal, real))*100;
Do you have any objections?
Одна и та же ТС при разных параметрах будет иметь разный КПД, и, естественно стоит выбрать параметры с большим КПД. Однако, такой показатель не имеет ни чего общего о доходности рассматриваемой ТС и это другой аспект исследований - вопрос ММ.
Речь идет конечно о переворотных ТС c последовательно идущими противоположными по направлению транзакциями.
Yeah, something like that.
But this is only a tiny tip of the visible, and usually discussed, energy of the CD due to the discreteness of the decisions made by the TS. In fact it is much larger, and can be estimated by an incredibly large, but nevertheless finite number of differently directed simultaneously opened positions. And there is only one limitation - the limitation on the maximum number of simultaneously opened positions, imposed by the DC. Such a flow of numerous simultaneously opened positions would give a smooth, almost analogous "overflow" of the RC energy.
This is exactly not a problem. After all, we should not compare different TS, but different parameters of ONE TS in relation to the ideal ..... (I think there is no such a notion, but the word "profitability" does not fit here, as I wrote above) of the RC energy.
That's generally understandable too. Perhaps that is why the word "indicator" began to appear in this discussion - as a primary forecast signal that does not include MM or other tricks.
А зачем надо сравнивать ?
Программер, это наверняка ты придумал сравнивать, а зачем ?
And it says at the beginning of the thread - to develop modesty.
;)
А там написано в начале ветки - для развития скромности.
;)
ReadSo let the programmer take his own, well, the one he successfully trades by, not the zz, as a benchmark.
He will quickly transmit his signals here during the day and everyone will compare them by the end of the day and draw conclusions to develop his own modesty
Например, пока рыночная позиция скажем бай-1-лот - индикатор тянет линию ==+1, затем закрыли бай == 0, открыли селл-2.3-лота == - 2.3 и т.д.и.т.п.
В таком варианте сигнализации для измерения КПД подойдёт корреляция Пирсона. Второй вариант (очевидный) - тестер. Но он требует двух прогонов и последующих расчётов.
А тут на выходе - сразу КПД по формуле кпд = abs(корреляция(идеал, реал))*100;
Возражения есть?
Yeah, I see. OK, let's move on.
Here's a piece of history - let's say 15 bars. The first 5 bars the price goes up, then 3 bars go down, and the remaining 7 bars go up. The ideal one (WP) shows +1 on 15 bars (that is how the WP parameter has been selected), i.e. just buy.
The real one opens three new positions. The signals are as follows:
-1,-1 (this is from the position opened earlier),
+1,+1,+1,+1 (we have finally seen the price go up),
-1,-1,-1 (a sharp drop down, but a delayed reaction),
+1, +1, +1, +1, +1 (reaction with a delay to the rise of the price).
Let's calculate the efficiency according to your formula. It turns out to be less than 100, of course. But how does this result relate to profitability, if the ideal system yielded 100 points on these 15 bars, and the real one - 110 (it may well be, because the real one has reacted to all basic price movements on 15 bars, while the ideal one saw the entire movement as a whole and stupidly opened buy)?
Да и сам-то вопрос был, собственно, к MD. Ну вот теперь еще и к Candid'у.
Понятно, что эффективность системы "две машки" не идет ни в какое сравнение с ЗЗ-идеалом. Боюсь, вообще никакая реальная система этого не обеспечит. Я все время пытаюсь протолкнуть идею о том, что ("исследуемой ТС").
Или пытаться оценивать систему так, как я предложил ранее, - через качество входа и качество выхода для каждой позиции и затем уже общее качество всей системы. Но тут и идеал не нужно строить: он как бы виден сам собой и полностью соответствует самой иТС.
In fact, I also believe that any realistic system will have negligible efficiency, no matter how you calculate it. That's why I don't see much point in calculating it. The only reason I got in is because you asked a specific question. And why do you think that my proposed algorithm of construction does not meet your condition: "the ideal itself must be built with the specifics of the ITS in mind"?
As for me, nothing characterizes the TS like its equity curve :) .
..........
.... Но какое отношение этот кпд имеет к финрезультату, если идеальная система на этих 15 барах принесла 100 пунктов, а реальная - 110 (такое вполне может быть, т.к. реальная смогла отреагировать на все основные движения цены на 15 барах, а идеальная увидела все движение вкупе и тупо открыла бай)?
An ideal-profit zigzag can be constructed. It is calculated using the formula H=Spread+1pips. It can be taken as 100% efficiency. And you can't beat it.
// This is to the question "about ideals" and "standard tests on standard ideals".
But such a test can come in handy. Purely to calculate some kind of standard "zig-zag" (c) trading system.
:)