Optimisation and Out-of-Sample Testing. - page 8
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
Surely there is already a practical implementation of this algorithm... On the forum I found only its derivatives... For example, 'How to implement your optimization criterion'...
I want to share my solution to this problem....
Let's prepare an EA... Let's add external parameters...
Into function init() insert the following block....
Parametr1, Parametr2, Parametr3 - external parameters, which should be optimized....
That's all there is to it...
You can read more about it here http://www.fxexpert.ru/forum/index.php?s=f657fc83fb442cf1fa2afde6fd4c37c2&showtopic=1106 post DM_35 Attached are pictures and everything you need.
You can read more about it here http://www.fxexpert.ru/forum/index.php?s=f657fc83fb442cf1fa2afde6fd4c37c2&showtopic=1106 post DM_35 Attached are pictures and everything you need.
>> Thank you. I guess my question to DolSergon is removed ..... though it's interesting too. Sometimes it is very useful to look at it with "hands and eyes". Very often, promising patterns, and not promising logic in an expert, are captured by "eyeballs" .... :)
Thank you. I guess my question to DolSergon is removed ..... though it's interesting too. Sometimes it is very useful to look at the whole thing "with hands and eyes". Very often, promising patterns, and not promising logic in an expert, are captured by "eyeballs" .... :)
It's useful, but you can't look through 2000-3000 results in manual mode.
Yes, no doubt useful, but you can't look through 2000-3000 results in manual mode.
That's the way I want it, so that after three or five periods and, consequently, "purges", 20-30 options are left..... it's already realistic to look at.
Well, let's wait, maybe he will answer :)))
Slightly off-topic, but mostly on-topic.
The proposed method of sequential optimisation can be performed on the same piece of history,
according to different criteria.
As a result, we will get a sample that contains variants optimized by all selected criteria.
Slightly off-topic, but mostly on-topic.
Checked it several times. Most of the time the same options are displayed, just sorted differently.
Almost 24 hours have passed since I visited the link from CtFelix.
More details about it you can read here http://www.fxexpert.ru/forum/index.php?s=f657fc83fb442cf1fa2afde6fd4c37c2&showtopic=1106 post DM_35 Pictures and all needed are attached.
Tried to use sequential optimization. Even if you do not take into account the technique and a lot of routine manual operations, the result is deplorable - by the end of the third-fourth run of the set of options, as a rule, nothing remains and I would not say that very much time is saved, especially when you test for all ticks.
By the way, it's interesting that most of the options on the 2007 run are cut off - what makes it so special?
There, on the link, one more approach is considered quite opposite: short-term trend fixing but I don't really believe in such a methodology.
But it seems to me that Vita is right when saying that optimization should be performed over the whole set of data.
You just need to flexibly use "optimization" tab in properties of the Expert Advisor - it also saves machine time, by the way. For example. If I need to limit the drawdown (not the percentage) and I don't see it in the tab, then I put a deposit of 10000000 and the drawdown percentage, which gives me the required value/amount. The trick is that for 10000000 or 10010000 0,02% is approximately the same, but the 20% drawdown for 10000 and 15000 is completely different order..... can someone else share some other tricks? )
Besides such optimization will not give many variants and if the expert is not "loaded" with anything positive, it will not give any variants at all.... . :)
If you really want, you can leave a bit of history and forward using the sequential method to finally verify the workability of parameters and then optimize the selected variant to the end. But again, there are no guarantees that it will work in the future).
Question. An EA with deep stops works in such a way that several uncompensated orders with significant current losses may be opened at certain time intervals and then, in the majority of cases, they would cumulatively turn to the profit through order management. If this is the time period when optimization finishes, all of them will be closed with the current loss. At the same time, we can see that the balance is sharply increasing towards equity at the end of the chart. Of course, if there is a drawdown limit, this variant is discarded.
How can this be avoided?
well, at least "meow" :)
>> meow :)
If you use Vita variant then you may get data fitting and I doubt anything good will come out of it.
As for taking too much time, I think it's more about the Expert Advisor's code or a large amount of optimized data, rather than the optimizing loop's code, which makes the optimization take a long time.
Question. An Expert Advisor with deep stops works in such a way that several uncompensated orders with a considerable current loss can be opened at certain time intervals and then, in the majority of cases, they cumulatively return profit through order management. If this is the time period when optimization finishes, all of them will be closed with the current loss. At the same time, we can see that the balance is sharply increasing towards equity at the end of the chart. Of course, if there is a drawdown limit, this variant is rejected.
How it may be avoided?
And here we should probably either remove the piece of history so it doesn't open this row of positions or add a piece of history so it can close them as it should.
And here we should probably either remove the piece of history so it doesn't open this row of positions or add a piece of history so it can close them as it should.
Or better rewrite that bit of history to make the Expert Advisor more comfortable. :))