"The 'perfect' trading system - page 95

 
Mathemat >> :

OK, let's get started. Actually, it is probably impossible to build a perfect EA. But first I will list its properties as I understand them. And then let's move on to the quasi-ideal as some approximation of the ideal.

1. Does not have a single arbitrary parameter that is not justified logically.

2. Always trades in profit, i.e. no balance drawdowns.

3. Each transaction is accompanied by the price change only in profit, i.e. there is no drawdown of equity below the balance. The price, already after the transaction has gained some paper profit, can go against the transaction, but never making the transaction unprofitable.

4. The previous point has a corollary: the strategy remains profitable even with an extreme MM.

I'll stop for now as it's time to go to bed.


My suggestion:

1. the trading robot must have a stable "carrier".

Everything else you listed can be achieved over almost the entire period of the price series, excluding periods of instability.

For example, if a sharp and long trend started, and the system happened to be in a phase - "against the trend" - this is the period of instability - you need time to stabilize.

 
VictorArt >> :


Have you read the beginning of this thread? :)

You announced the criterion - I just pointed out that the Adaptive EA falls under this criterion.


" Angela wrote >>.

1). One of the most important principles underlying ITS I believe is its ability to self-learn and adapt to changing market phases. And the second, 2). The minimum number of externally adjustable parameters.

At the link, there is an adaptive algorithm that satisfies your criteria:
1. ability to learn and adapt
2. only 1 optimizable parameter

Yes, I remember, while your Expert Advisor does not meet the most important principle - "stability of profit growth".

All other benefits do not count until the main condition is met.

 
VictorArt >> :


Well, the term "stability" which has been "laughed at" so much has come in handy :)

Sorry, but the terms "maximise stability" or "minimise risk stability" make no sense.

The main parameters of the system are "stability of profit growth", or the associated "risk minimisation"

 
Pegasmaster >> :

Yes I remember, as long as your EA does not satisfy the most important principle - "stability of profit growth".

All other benefits do not count until the main condition is achieved.



It is not the main principle, but a secondary one.

A stable carrier is more important.

Carrying:

Equity:

 
can we stop selling air already?
 
VictorArt писал(а) >>

This is not the main principle, but a secondary one.

A stable carrier is more important.

Carrier:

Equity:

I've written to you many times before - you have a problem with over-optimisation, so there is a drain. By the criteria you are optimising against, this is what leads to over-optimisation.
 
Pegasmaster >> :

Sorry, but the terms "maximise stability" or "minimise risk stability" make no sense.

The main parameters of the system are "stability of profit growth", or the accompanying "risk minimisation" thesis.



When it comes to 'stable carrier', profit is irrelevant - see pictures above.
 
VictorArt писал(а) >> profit is irrelevant.

Back to anal. Where's the anal branch?...))))

 
VictorArt >> :


When it comes to "stable carrier", profit makes no difference - see pictures above.

>> it matters how many suckers the advisor buys

 
VictorArt >> :


My option:

1. the trading robot must have a stable "carrier"

Everything else that you listed can be achieved for almost the entire period of the price series, excluding periods of instability.

For example, if a sharp and long trend started, and the system happened to be in a phase - "against the trend" - this is the period of instability - you need time to stabilize.



1. Your variant has a right to life, but use common notions, or explain what you mean, because "it must have a stable "carrier"" seems to be understood only by you.

2. If the system is "perfect" enough, it cannot "happen to be in the phase - "against the trend"", unless its behavior was originally designed to do so

This is exactly the kind of system we are discussing. "Accidentally" is not about such systems.

p.s. Don't be distracted by your advisor, maybe we'll learn something new