How do I check whether 'Optimisation' or 'Forward Optimisation' is in progress? - page 9

[Igor Volodin]

Youri Tarshecki:

I see that you have sacrificed transaction density in favour of timeline - don't know if that's right.-)

Before I did it without time scale, it wasn't very clear. Uniform profits and losses.
If my Expert Advisor is attached to a robot's timeline, then it will visually identify the periods of robot's operation and inactivity. Otherwise we cannot filter out sets that start trading once in a six month period at some sideways trend, look attractive and profitable, and then remain silent.

[Igor Volodin]

Dmitry Fedoseev:
Where do I get the number?

while(FrameNext(pass,name,id,value,data)) {
   bool is_forward = CheckPass(pass); //проверим pass в пуле
   if (!is_forward) {
     AddPass(pass); //добавим pass в пул

     //обработка бэка
   } else {

     //обработка форварда
   }
}//endwhile

[Dmitry Fedoseev]

Igor Volodin:
Are you calling me a liar?

No, a taciturn one.

[Youri Tarshecki]

Igor Volodin:
I used to do it without timing, but it is not very obvious. Uniform profits and losses.
Drawing it in accordance with time allows you to visually determine the periods of robot operation and downtime. Otherwise we cannot eliminate sets that start trading once in a six month period and look attractive and profitable, but keep quiet after that.

After thinking about it for a while, I agree. But here are the starting deposits, it seems to me, should be from the same point. Well, or to provide for both variants.

And also, it will be very useful if it is possible to load volking forward runs. Is it technically possible?

[Dmitry Fedoseev]

Igor Volodin:

Thank you!

[Igor Volodin]

Youri Tarshecki:

forwarking-forwalking-forwards. Is this technically possible somehow?

In fact, it is less important than looking for sets of suitable parameters on history. In the end the rolling forward will give (filter) a set of parameters giving the balance curve with acceptable values at every next period. And it can be searched in another way.

We already discussed custom optimization criterion here - https://www.mql5.com/ru/forum/74809#comment_2301588

After backtest with the mentioned criterion we check all found balance curves with the last forward and look through all outcomes to avoid cases when parameters are fitted to the result.

[Youri Tarshecki]

Igor Volodin:

In fact, it is less important than looking for sets of suitable parameters on history. In the end, the rolling forward will still give (filter out) a set of parameters giving the balance curve with acceptable values for each next period. And it can be searched in another way.

After backtest with specified criterion, we check all found balance curves by last forward, looking through all outcomes to avoid cases when parameters are fitted to the result.

The question is how to analyse the RESULTS of a volking forward. And its results are those single back-and-forth runs that occur after back optimization. I.e. the single run with the set that is the coolest according to the user. Suppose we have 12 wolfing-forward runs.

Question - how should we assemble these 12 runs or what should we do to cram these runs into such a run?

[Forester]

Youri Tarshecki:

It's about how to analyse the RESULTS of a volking forward. And its results are those single back-and-forth runs which occur after back optimization. I.e. the single run with the set that is the coolest according to the user. Suppose we have 12 wolfing-forward runs.

Question - how should we assemble these 12 runs or what should we do to cram these runs into such a run?

I manually exported the optimization results to XML, then opened the file in Excel and exported it to a delimited text file. Then I exported it to MySQL database. In the database I was able to sort, filter, etc. Although, this can be done directly in Excel. DB is more convenient because you can put there the results of many runs, adding the appropriate labels.

[Youri Tarshecki]

elibrarius:
I manually exported optimization results to XML, then opened file in Excel and exported it to a delimited text file. Then I exported it to MySQL database. In the database I was able to sort, filter, etc. Although, this can be done directly in Excel. DB is more convenient because you can put there the results of many runs, adding the appropriate labels.

I also automatically pull the tester's report data into Excel and get the run pictures by copy-paste, but they are numerical values and separate pictures.

I would like to see a chart with all 12 equities ONCE in it. And I want all forwards to start from the same point.

[Forester]

Youri Tarshecki:

I also automatically pull the tester's report data into Excel and get the run pictures by copy-paste, but they are numerical values and separate pictures.

But I would like to have a graph with all 12 equities ONCE. I also would like all forwards to start from the same point.

For me, numerical values of forward (profit and drawdown) are enough so far. Charts are great, of course, but they require saving data of all trades and subsequent creation of equity on them. I think the development costs will be too high, while the advantage is only the clarity.

Besides, trading operations data can be saved in files or databases only when optimizing on your computer. When optimised in the cloud, we can only get a standard report.