From theory to practice - page 332
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
It has nothing to do with plots, he did it all in real time.
The chances of it being an accident are slim to none.
Not entirely negligible... If you happen to find a good trend, it may work, but you don't know if it will work in other areas.
It has nothing to do with plots, he did it all in real time.
The probability of it being an accident is negligible.
No matter how you adjust parameters in the tester, the forward will show what works and what does not.
The problem with the tester is that forward becomes a part of optimization if you repeatedly defect the TS.
You yourself actually become a GA element that is doing the intelligent rejection.
But while GAs somehow try to protect themselves from overoptimization, no one is going to control themselves.
That's why the real-time state is the last frontier of testing. In realtime, any tweaking will turn out to be detrimental.
Quite right, Nikolai. Optimization and tweaking will give good results only on the optimized area to show good results in the tester and make lots of suckers by selling such EAs.
It's just that you can see that almost all trades are to one side and you can see a strong trend section so I wonder how it would work on a flat for example...
The tools are different. Do you understand that?
The tools are different. Do you understand that?
There's a strong trend section just there, it usually correlates on other instruments as well.
Nikolay Demko:
So the real-time stats are the last frontier of validation.
There is such a thing as statistical validity... If it is too short, it may work fine, but a little longer than BP, the statistics may be different...
There is such a thing as statistical reliability. If the time frame is too short, it may work fine, but a little longer than BP, the statistics may be different...
Statistical validity is calculated in the tester, when developing TS.
To make sure the TS is good enough (after the completion of work in the tester) it passed only two parts in real time, a trend part and a flat part.
There is a strong trend segment there and it usually correlates with other symbols as well.
:))
And on TF 5 minutes I always have trend sections.
No, I sit and choose Mondays with trend sections after Sunday's rubbing). Here is another example from Monday#62
Mondays should be banned).:))
And on TF 5 minutes I always have trend sections.
No, I sit and choose Mondays with trend sections after Sunday's rubbing). Here is another example from Monday#62
Mondays should be banned).To make sure the TS is good enough (after completing the work in the tester) it has passed only two sections in real time, the trend section and the flat section.
Well, yes, on a flat, we should also see if it works...
Statistical validity is calculated in the tester, when developing TS.
To make sure the TS is good enough (after the work in the tester is finished) it has passed only two parts in real time, the trend part and the flat part.
Isn't it you?
Nikolai Semko