Optimise an EA and get the best of the optimised ones. - page 34
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
Another TC for over-optimisation:
Update of 16:40, 21.04.18:
Removed over-optimized system from the table.
Taking EURAUD EMATrendRTS
Optimisation period 21.04.17-21.04.18 forward from 21.09.17
I understand.
But, I paid for this code (or rather, not even for the code, it is not complicated there, but for the idea itself), and I do not want to disclose it to a wide circle. So, friends, excuse me, there will be no "quality" formula. I see that this indicator - very adequately reflects the nature of the balance curve, in fact, the quality of trade, which is why I called it so, and which is why I abandoned all other evaluation criteria in favour of this one. I don't use any other criteria myself. Personally, Alexey, I can tell you the gist in private.
As I have repeatedly said, the only thing I lack is some kind of indicator of TC stability. Quality - it reflects the "degree of niceness" of the curve. This is a good indicator, but no less important is the ability of the TS to "resist" small changes in the market. And here I, alas, have "zero", except for vague reasoning and intuition - nothing. If I manage to formalize this figure - then I won't need anything else for selecting systems. We set a "threshold of stability", sift out all TS, which have it below, and select the highest quality of the remaining ones.
Anyone can get the usual information on TC without any problems - the magician is known, and the main working file of the TC League is available. We download it, indicate the desired TS, and watch the drawdown, recovery, profitability and everything else.
No, well, that's no way to sell a product - if you know it's good, it doesn't mean that everyone will believe you. If you do not want to reveal the secrets, it's better not to show this parameter here, because it is not available to all, and show those that are standard - I do not think it is so difficult to calculate ... Well, maybe exclude the drawdown by means - and it is not necessary, if the work is done with a stop.
Think about the public's interest in the project - you need an adequate presentation that can interest without unnecessary gestures, and for this you need to present information in a familiar way.
And think about probability....
No, well, that's no way to sell a product - just because you know it's good doesn't mean everyone will believe you.
Nah... Alexei, I don't feel like selling anything. I even open a signal from the real with great reluctance - if it were not for persistent requests - I would not do it at all.
I've only got four private messages here and three people wanted to see the real-time signal from the best TS in the league. And on another forum, where there are few programmers and a lot more people - I've already had about two dozen people say that "if you do not have a signal, or at least a real-time chat - then you're a scammer, and you want to rip us off".
I said I wanted to create a kind of community, in which participants would be engaged in overoptimization of outsiders, and would use the TC favorites on a "non-monetary" basis. Money - we'll make money trading. But... This idea, as I understand it, is not interesting for everyone. Here the interest was expressed by ten people, the real help came only from you and one more. And on the other forum - one person had reoptimized five systems, and now he is watching. Plus another one - all the "pricey", going to sort of optimize something, but has not decided yet. And the vast majority want nothing less thana "Grail with real estate guarantees". Putting an already idle computer on for a couple of hours is a terrible strain for them. It's even been stated that I de "want to use the free slave labour of the newbie losers".
So - there will be a signal. People want to pay money, but that their computers are not loaded - fine, let it be so. Although, I don't really like this idea.
And think about probability....
Yes, this is an important area...
Nah... Alexei, I don't feel like selling anything. I'm not even willing to open a signal from the reals, if it were not for the persistent requests, I wouldn't do it at all.
I've only got four private messages here and three people wanted to see the real-time signal from the best TS in the league. And on another forum, where there are few programmers and a lot more people - I've already had about two dozen people say that "if you do not have a signal, or at least a real-time chat - then you're a scammer, and you want to rip us off".
I said I wanted to create a kind of community, in which participants would be engaged in overoptimization of outsiders, and would use the TC favorites on a "non-monetary" basis. Money - we'll make money trading. But... This idea, as I understand it, is not interesting for everyone. Here the interest was expressed by ten people, the real help came only from you and one more. And on the other forum - one person had reoptimized five systems, and now he is watching. Plus another one - all the "pricey", going to sort of optimize something, but has not decided yet. And the vast majority want nothing less than a "Grail with real estate guarantees". Putting an already idle computer on for a couple of hours is a terrible strain for them. It has even been stated that I de "want to use the free slave labour of newbie losers".
So - the signal will be. People want to pay money but that their computers are not loaded - well, let it be so. Although, I do not really like this idea.
To sell, in that context, is tantamount to "interest". Interest in numbers is more stimulating than an idea in words. And right now, numbers aren't enough. Yes, of course you can go digging yourself, but, what's the incentive for a person to do that?
In fact, I think that 5-10 people are enough for this project, it is judging by the dynamics of the need for overoptimization.
For them and you can open a free signal on a cent account. Then people will have an incentive to optimize to improve/support the signal.
But few will bother with codes and other stuff from optimization.
I would connect to a free signal, just for the sake of interest.
To sell, in that context, is tantamount to 'interest'. Interest in numbers is more stimulating than an idea in words. And right now, numbers are not enough. Yes, of course, you can go digging around on your own, but what incentive is there for a person to do that?
Well, in theory, in the coming weeks - I can do it, add columns to the report, and the tables will add fields "drawdown", "recovery", and others (although, the maximum drawdown - in the table already has, in addition - there is a maximum queue SL). This data is calculated inside the system, you just need to add it to the report. Although, in my opinion, report tables are already a bit long. Add more columns ? Is it worth it ?
Especially, now I show the top ten TCs, and I look that it would be good to show all TCs with quality above 80% and their - three times more ! Are you sure that people will understand this endless number of tables?
Isn't the view of the balance curve plus some kind of estimate that reflects its quality well enough ? Especially since it is a curve constructed not in a tester, but on the basis of real trades on a demo account !
In fact, I think 5-10 people are enough for this project, it's judging by the dynamics of the need to over-optimise.
For them and you can open a free signal on a cent account. Then people will have an incentive to optimise to improve/support the signal.
But few will bother with codes and other stuff from optimization.
I connect to a free signal, just for the sake of interest.
Yes, in principle, the assessment is correct. Ahead of hundreds of three new TC, well, reoptimization will require an average of three to five pieces daily.
About the signal... It seems to me, the signal is needed by people who just do not bother with anything. For them, a paid signal would be preferable.
And those who participate in the project - isn't it easier to put the best (in their opinion) of the League's TS on at least a demo, at least on the real account? And have a signal, and trade, at least a demo, at least a real one? I don't quite understand, why do you need a signal with your pile of rekords? You can set all favorites for the whole year! In the next month I'll try to add to the League the ability to read regexes from .ini file, then you can run even a dozen of TS by one Expert Advisor at once ...
Launched
Well, in the next few weeks, I can do this, add columns to the report, and add "drawdown", "recovery", and other fields to the tables (although the maximum drawdown is already in the table, and there is also a maximum SL queue). This data is calculated inside the system, you just need to add it to the report. Although, in my opinion, report tables are already a bit long. Add more columns ? Is it worth it ?
Besides, I'm showing the top ten TSs now, while I see that it would be better to show all TSs with the quality higher than 80% and there are three times more of them ! Are you sure that people will understand this endless number of tables?
Isn't the view of the balance curve plus some kind of estimate that reflects its quality well enough ? Especially since it is a curve constructed not in a tester, but on the basis of real trades on a demo account !
It is worth adding as long as the width allows :)
It may be good to show everyone, but it's not realistic on one graph - there is already a lot of noise... A couple of charts is better, or a slow scrolling gif.
The balance curve might tell me something, but I usually look at it after evaluating the numbers. By the way, it would be good to add an indicator of the average number of profitable and losing trades and an indicator that would reflect how many profitable average trades are needed to cover one losing one.
Yes, in principle the estimate is correct. There are still hundreds of three new TC's ahead, well over-optimization will require an average of three to five every day.
About the signal... It seems to me, the signal is needed by people who just do not bother with anything. For them, a paid signal would be preferable.
And those who participate in the project - isn't it easier to put the best (in their opinion) of the League's TS on at least a demo, at least on the real account? And have a signal, and trade, at least a demo, at least a real one? I don't quite understand, why do you need a signal with your pile of rekords? You can set all favorites for the whole year! In the next month I'll try to add to the League the ability to read regexes from the .ini file, then you can run a dozen TS at once with one Expert Advisor ...
If you're about to start, you need to put a limit on the number of pips that you want to give the pins to, and then you'll have to put a limit on the number of pips that you want to give.
The thing is, I don't want to bother with the selection theme and other stuff - I have enough problems of my own, and you do your TS every day, you know their pros and cons, and therefore can, with greater probability, select the best option and run a basket to work, because you're doing it for yourself. In general, I feel more comfortable being a hamster - I did what was required, and bam, I got a reward in the form of a trading signal. I think many people have such a logic. I.e. optimization is the price of the signal, but if you don't do it, the signal will be worse. And then everything depends on the signal. I mean, your idea that everyone will take responsibility for finding the right solution is, in my opinion, utopian - people need either a ready-made set or a ready-made signal. Signal is more convenient in terms of loading the PC with calculations.
Launched