The "Maybe we'll get lucky" counsellor - page 5
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
The irony is not understood.
Applying the theme of the topic of the topic, made your Expert Advisor profitable against your logic, and it turns out that "you obviously don't have enough basic knowledge to understand even such a simple question" is to blame.
Imagine, , that you tossed a coin every day while changing your jacket.
You soon realised that in the sequence in which you were wearing red, then blue, green, blue, purple, your jacket, you got tails.
You go to your buddy and argue with him that now being dressed in a green jacket you will get tails again.
The optimiser of the EA does the same thing, it finds a random parameter at which you get a profit, and it is absurd to believe that you will continue to have a profit based on the current parameters of the EA.
Imagine, , that you tossed a coin every day while changing your jacket.
You soon realised that in the sequence in which you were wearing red, then blue, green, blue, purple, your jacket, you got tails.
You go to your buddy and argue with him that now being dressed in a green jacket you will get tails again.
The EA optimizer does the same thing, it finds random parameters where you made a profit, and it is absurd to believe that you will continue to make a profit based on the current EA parameters.
You can write off even more, just calculate how many variants an EA can have when optimizing stops and profits
( tens of thousands)
That's a lot more variety of jackets on the planet.
But do 61 profitable and only 6 losing trades during a year tell us anything or not? Can a tenfold superiority be blamed on variance?
This suggests a terribly unequal distribution in the tails (base of the bell) region. That is, low probability density (tail) events have a much higher probability than in the normal distribution, at the cost of a slight reduction in the more likely events (the top of the bell) in the same distribution. As a result, three sigmas is not the limit.
This is one of the main reasons for "successful" fitting to historical data.
P.S. With a normal distribution, the probability of three sigmas is negligibly small: 0.0027
I leave this academic opportunity to you.
What am I, a head wounded man?
The sample size is too small. This 61 to 6 ratio could well turn into 40 to 27, for example. And if we take into account possible situation when SL ~ 2*TP, we may get profit factor less than 1. And that's roughly what the picture shows (if not worse):
But 610 to 60 is already a very serious bid, even with the current ratio of SL to TP.
The sample size is too small. This 61 to 6 ratio could well turn into 40 to 27, for example. And if we consider a possible situation where SL > TP, we might get a profit factor of less than 1.
Here 610 to 60 (with commensurable SL and TP) is already a serious bid.
If we assume that all passes lead to about 700 trades.
We have 10 000 passes of optimization.
What are the odds that there will not be a single option at which the 640 to 60 ratio is met?
Assuming that all passes produce around 700 trades.
We have 10,000 passes of optimisation.
What is the probability that there is not a single option where the 640 to 60 ratio is met?
The problem conditions are not sufficient to get a solution. You have not given the frequencies at 700 trades.
And the main question is not very practical: exactly 640 to 60 - or anything better than that ratio?
The conditions of the problem are insufficient to get a solution. You have not given the frequencies at 700 transactions.
For simplicity, tp and sl fall out with equal frequency.
Yes and the main question is not very practical: exactly 640 by 60 - or anything better than that ratio?
Anything better also satisfies