Combining multiple indicators in an EA - page 2

 

This mathematics is worthless at all, as it does not take into account the relationship between the signals.

 
What do you need them for, that 50%. 50% is a lie. And 20% is a big lie. And 30% is the right thing to do.
 
Mathemat писал(а) >>

This maths is worthless at all as it does not take into account the relationship between the signals.

I absolutely agree that it does not take into account, but it should.

 

Still, what is the 'right' signal? All indicators give the "right" signal. But for some reason, the result is usually not what you want to get. I will say more. All indicators ALWAYS give the "right" signals.

Even if you enter without any indicators, you can make a profit 99% of the time!

The question is where to place the stops. But there's nothing about the stops here.

 
joo писал(а) >>

Even if you go in from scratch with no indicators at all, you can make a profit 99% of the time!

It's not really about the stops. The question is how much profit you will make and over what period of time. And this

"ratio may not suit you, to put it mildly.

A signal that provides the following ratio can be considered as a correct one

(order work period) and the drawdown size you like.

 
Richie >>:

Не совсем в стопах. Вопрос в том, сколько вы прибыли получите и через какой промежуток времени. И это

"соотношение" вас может и не устроить мягко говоря.

Правильным сигналом можно считать сигнал, который обеспечивает такое соотношение

(размер прибыли \ срок работы ордера) и такой размер просадки, которые вас устраивают.

That's what I'm talking about. And it's all about the stops. Depending on where they are set will determine the correctness of the signal.

PS Therefore, as Mathemat correctly pointed out, the approach which calculates the probabilities of determining the correctness of signals without taking into account their mutual relationship with each other would not be correct

 
joo писал(а) >>

That's what I'm talking about. And it's all about the stops. Depending on where they are set will determine the correctness of the signal.

PS Therefore, as correctly noted by Mathemat, the approach that calculates the probabilities of determining the correctness of signals without taking into account their mutual relationship with each other would not be correct

Of course one must take into account, I absolutely agree with respected Mathemat. But, this mathematics I cited is not for professional mathematicians,

It's for beginners who believe that it's possible to make one useful indicator from 10 useless ones that everyone uses.

Let's get 20 schoolchildren together in a pile. Can they design an intercontinental missile? No. You need a smart head for that, at least one.

And as for mutual influence - here lies another rookie mistake: another attempt to cross a few wizards or stochastics.

 
avatara >>:

если перевести "правильность" в вероятность - то умножение их "сигналов", что нам покажет?

;)

один врёт с вероятностью 50%, второй - 40%.

Вместе врут на 20%.

:)))))

Но - правильно предсказывают 0,5х0,6 =0,30.

Итого правильно предсказывают на 30%.

---

Разве не так?

только 50% куда то делось... :(

The "gurus" have brushed it off... :(

But - 0.5x0.4=0.2 the first one said it correctly and the second one didn't.

0.5x0.6=0.3 the first one lied...

:о)))

50% interfered with each other!

Or what?

;)

It turns out that if we believe all positive signals - the probability of luck is already 30+50=80%.

The grail is near.

 

avatara,

According to the "AND" principle: if the 1st indicator gives 50% correct signals and the 2nd only 60% correct signals,

then both of them, according to this principle, will give only 30% correct signals. At the same time, 70% of the remaining

signals will be incorrect. These 70% are the situations when one indicator gives wrong signals

signal and those situations where both give wrong signals. Both indicators will give wrong signals only in

signals in only 20% of cases. The correlation between the indicators is not considered at all.

 
Richie >>:

avatara,

По принципу "И": если 1-й индикатор даёт 50% правильных сигналов, а второй только 60% правильных,

то оба они по этому принципу будут давать только 30% правильных сигналов. При этом 70% остальных

сигналов будут неправильными. Эти 70% - это и те ситуации когда один индикатор даёт неправильный

сигнал, и те ситуации, когда оба дают неправильный сигнал. Оба индикатора будут давать неправильные

сигналы только в 20% случаев. Взаимосвязь между индикаторами не учитывается вообще.

Let's assume there is no multicolinearity.

30% is both. And apart, it's another 50%.

What's next, Master?

;)