Trader's self-deception: distrust of forwards. - page 16

New comment
 
Youri Tarshecki:
The hyphoto already shows it all. Step by step. Step 1, Step2...which means Step 1, Step 2...Once again, please don't litter my thread if you don't think wikipedia and Amitrade, who even built Walk-Forward into their platform, are authoritative - I have nothing to do with it.
Deal. Adiós Carlo! Z.U. There aren't many people on here who would defiantly ignore even my writing.
 
Youri Tarshecki:
The hyphoto already shows it all. Step by step. Step 1, Step2...which means Step 1, Step2 ..Once again, please don't litter my thread, if you don't trust wikipedia and the manufacturer Amitrade, who even built Walk-Forward into their platform, I have nothing to do with it.
you can make the start of the test as wholesale and optimise in 1 month increments (no GA). The back test period is fixed. Then the optimization results will contain all results separately for each backtest zone, the best ones will be automatically saved to a file at the end of optimization. Then, when running in the forward mode, depending on the current date, the Expert Advisor will load the necessary values of opts from the file. Total 2 times the Expert Advisor is launched)
 
Avals:
it is possible to make the start of testing as an opt-in and optimise in 1 month increments (without GA). Backtest period is fixed. Then the optimisation results will have all results separately for each backtest zone, the best of which are automatically saved to a file at the end of optimisation. Then, when running in the forward mode, depending on the current date, the Expert Advisor will load the necessary values of opts from the file. In total, the Expert Advisor will be launched twice)
Another good suggestion, but the disadvantage is that the tester will run in idle mode most of the time.
 

Sometimes, or very often, some traders blindly believe in certain strategies. Without any logical reason. Next step for them, writing an EA on their chosen strategy, and surprise.... It plummets. But still, the trader continues to blindly believe in the chosen strategy, and then it starts.... Optimization, forwards, formulas, and other tricks of self-deception.

Thus, wasting a lot of time, instead of spending it profitably. Sinks, moves on to another....

PS Strategy is not a child to be "brought up".
 
Ghenadie Tumco:

Sometimes, or very often, some traders blindly believe in certain strategies. Without any logical reason. Next step for them, writing an EA on their chosen strategy, and surprise.... It plummets. But still, the trader continues to blindly believe in the chosen strategy, and then it starts.... Optimization, forwards, formulas and other tricks of self-deception.

Thus, wasting a lot of time, instead of spending it profitably. Sinks, moves on to another....

PS Strategy is not a child to be "brought up".
And how do you check if the strategy is losing all the time or it was an isolated incident. A child may screw up today and be a good boy tomorrow. Strategies do not educate, they correct.
 
Ghenadie Tumco:

Sometimes, or very often, some traders blindly believe in certain strategies. Without any logical reason. Next step for them, writing an EA on their chosen strategy, and surprise.... It plummets. But still, the trader continues to blindly believe in the chosen strategy, and then it starts.... Optimization, forwards, formulas and other tricks of self-deception.

Thus, wasting a lot of time, instead of spending it profitably. Sinks, moves on to another....

PS Strategy is not a child to be "brought up".
Forward analysis is precisely designed to objectively check the strategy and, if necessary, change it. I'm talking here about automating the process of changing an EA based on forwards. But the favourite tool for self-deception is fitting of all sorts, not forward.
 
Youri Tarshecki:
I'm talking about automating the process of changing an EA based on forwards.
How does that work? The robot writes the robot?
 
Yuri Evseenkov:
How's that? A robot writes a robot?
Approximately. You can start by automatically selecting the best variant from the ones uploaded for testing. And then it won't be long before the autotester generates the variants itself.
 
Youri Tarshecki:
The Autotester will then be able to generate variants on its own.

This is all sophisticated curvafitting. If such a theme were profitable, you would have to compete with a bunch of quants with super resources from large foundations. No fish there)) Unless you create artificial intelligence))

 
Youri Tarshecki:
Approx. To begin with, you can now automatically select the best variant from those uploaded for testing. And then the autotester will be able to independently generate variants on its own.
The task seems to me very resource-intensive. Do you have your own tester or do you use your own one?
1...91011121314151617
New comment