Determine the future operability of the vehicle. - page 5

New comment
 
LeoV писал (а) >>

Also understandable. But that's not exactly what I'm asking. These are the principles of building an un-optimized TS, preferably without optimization. And I'm trying to find out how to determine the capability of the TS to work in the future by using reports from the optimization period and from the OOS period.

You can't. There are standard market situations characterized by some number of parameters. When tested on a certain piece of history, we get a certain set of them with the same parameters. As a result of optimization, your Expert Advisor more or less successfully processes such a set. If market situations do not differ too much in CB, the test will be successful, if too, the result is unknown.

In order to know (rather than guess) an Expert Advisor's ability to trade successfully, you need to know which situations and at what parameters it can handle, and which ones it cannot.

And to do this, we need to test it not on history, but on these situations, and by varying its parameters, determine the area of its successful work.

 
If you take the tester as 100%, the demo will be 90%, the realisation is down to 50%
 
2LeoV . You've been on the subject for a long time. Don't you know that there is no such parameter, by the value of which, without knowing the system, you can say for sure about its successful perspectives from two tester reports.
As it was already said here, you can find out the robustness properties of a system, for example, by performing the multiple forward analysis.
Then we can say that with certain results in the past, the system is likely to behave similarly in the future.
 

Leonid, I've been thinking about it for a long time too - and I'm not the only one, obviously. And not only about "classic" Pardo testing, but also about alternative methods (there's an outline of one method in my article on sandwiches, but it probably won't work for you in that form). Perhaps I'll post something in the article. But it's a case of not doing it soon enough.

 

Con-Kop has an interesting software program on their website called 3-D Analyzer at http://konkop.narod.ru/.

It is very interesting to observe how the spot of optimal parameters moves with time.

Maybe someone will write a script for Omega or Metastock. It is very convenient for analysis.

 
ivandurak писал (а) >>

Con-Kop has an interesting software program on their website called 3-D Analyzer at http://konkop.narod.ru/.

It is very interesting to observe how the spot of optimal parameters moves with time.

Maybe someone will write a script for Omega or Metastock. Very handy for analysis.

>> Somewhere recently saw an implementation for MT, though more likely a manual.

 
In my opinion, testing and optimising on history is a waste of time. Except for intraday strategies, you open in the morning, you open in the afternoon, you close in the evening. As everyone knows, interest rates change all the time. When testing a long-term strategy, the tester generates swaps for the current moment in time. Of course, the testing and optimization result is distorted, as interest rates change very often. Try to test and optimize long and short positions separately. Most likely, many have done so and know the difference. I know there are brokerage companies that do not use swaps, but they do not work with MT and I do not really trust them. If anyone knows how to avoid this problem when testing and optimizing strategies please advise.
 
Vinin писал (а) >>

Saw an implementation for MT somewhere recently, though more of a manual

Now that's interesting. Where is it?

 
Serj_Che писал (а) >>
As everyone knows, interest rates change all the time. When testing a long term strategy, the tester gives out swaps at the current moment in time.
swaps are constantly changing? twice a year? and what kind of a system is it that depends on swaps for profit? or is it so "profitable" that it can't cover them?
 
olltrad писал (а) >>
what kind of system is it? does it have a profit based on swaps or is it so "profitable" it can't cover them?

a much better idea is to fight the spreads. somehow bypass them (does anyone have a script to deactivate the spread? :) )

12345678910
New comment