Looking for Yoda - page 2
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
There's a "school case" regarding neural network, it seems to me they call it : "byzantine fault tolerance" (to check!) - example, my EA is using actual datas of the footsie while british are preparing for the brexit (or no), once the brexit will be done (or no), the market won't be the same, it'll require to be updated. It's an easy to handle example as the brexit is a huge event.
Unfortunately I don't have a terminal installed at the moment, I would have show you the example of a stable EA since 01.2015 and probably before becoming unstable only in 07.2018 : God's creation, the devil's share ...
Of course not. In general, for the creative process should not be "should be"
Sorry, I meant if that idea would be viable or I am getting concepts wrong. The way I see it its like if I build a NN and I run an optimization every week to adjust weights, then why not adjust weights using GA.
There's a "school case" regarding neural network, it seems to me they call it : "byzantine fault tolerance" (to check!) - example, my EA is using actual datas of the footsie while british are preparing for the brexit (or no), once the brexit will be done (or no), the market won't be the same, it'll require to be updated. It's an easy to handle example as the brexit is a huge event.
Unfortunately I don't have a terminal installed at the moment, I would have show you the example of a stable EA since 01.2015 and probably before becoming unstable only in 07.2018 : God's creation, the devil's share ...
I see. I know markets change and evolve. That's why I'm looking to understand real optimization usages and how to improve it (using NN, GA or other ways).
I would love to see the example of the EA you mentioned. Why it became unstable at 07.2018? Anyways, 2015-2018 its quiet impressive I think. What do you do when an EA becomes unstable? You optimize it again? What period of time for back and forward testing should be considered to reoptimize an EA?
Sorry if they are so many questions, but I am pretty curious about this (still learning, I'm still a young padawan haha)
Regards,
PS
Yes but I won't reinstall a terminal, sorry dude.
I don't know why it turned instable, I suppose market conditions changed.
2015-2018 is little, many people will tell you they want at least 10 yrs of history but I don't have 10 yrs of historical datas.
You can reoptimize, make a correction if you can figure what's the problem .. or anything else that could solve it.
Yeah... you're aiming high for a padawan :)
Yes but I won't reinstall a terminal, sorry dude.
I don't know why it turned instable, I suppose market conditions changed.
2015-2018 is little, many people will tell you they want at least 10 yrs of history but I don't have 10 yrs of historical datas.
You can reoptimize, make a correction if you can figure what's the problem .. or anything else that could solve it.
Yeah... you're aiming high for a padawan :)
Haha thanks. I think I should aim high if I want to succeed at something. Just like if I am committed to do something, I really want to do my best!
By "terminal" do you mean the mt5 platform? Is there an EA that has 10 years of good performance? I think HUGE.
For example, how do banks and other institutions do this? I know of HFT executing orders in C ++, right on the server (like 1 million transactions in 1 second).
Can you tell me if I'm on the right path? If an NN that uses GA to obtain weights is a correct approach? Or I'm just dreaming and it's not possible / viable.
I would appreciate any advice you could give me for this trip (in addition to the other tips, which I'm already very grateful)
Regards,
PS.
Yes I meant the platform, not installed on this laptop.
Many EA on the market are optimized on 10 yrs, I can't provide no example as it's forbidden but you'll find easily.
Indeed in HFT there's thousands transactions per seconds, you can do it in MQL5 using the async mode, but you'll be limited by your hardware (connectivity, mem etc ...) - (personally I never done it so ... to check)
Not the right path, self optimization by GA just as for the tester ... but tell me how would you simulate a trailing stop like the tester do it to compute entry/exit/profit/pass result ? tick by tick ? come on ...
Not the right path, self optimization by GA just as for the tester ... but tell me how would you simulate a trailing stop like the tester do it to compute entry/exit/profit/pass result ? tick by tick ? come on ...
Honestly, I don't know yet. This is new for me, but nothing I can not learn for sure. Maybe I have the wrong concept for the use of GA?
I mean... lets do another approach to the question. What's the use of GA in an EA? What can I create? Is is for using in the OnTester() function that might can have?
Sorry, I meant if that idea would be viable or I am getting concepts wrong. The way I see it its like if I build a NN and I run an optimization every week to adjust weights, then why not adjust weights using GA.
The fact is that for a high-quality trading system, the analysis of the market situation and the auto-tuning of parameters depending on this situation should occur not once a week, but every tick. The trading system constantly analyzes such characteristics, for example, as amplitude, width and length of channels of all timeframes, spread expansion, movement strength, moment of breakdown of channels, etc. What many people call optimization, in fact, is a trivial adjustment of parameters to historical data. These parameters will be “optimized” only for this period of past history and this absolutely does not guarantee their success in the present and future.
Imagine that you are walking in the mountains and open your eyes for 5 seconds once an hour in order to memorize the terrain and change the speed of movement and direction. How far will you go of this method?
Imagine that you are walking in the mountains and open your eyes for 5 seconds once an hour in order to memorize the terrain and change the speed of movement and direction. How far will you go of this method?
Absolutely. Very nice example.
Absolutely. Very nice example.
Only a small clarification: you look not forward, but back. :))