Matringale or Grid-EA Optimized by AI or NN - does this exist?
I have seen some that handeled multi symbol hedging, so not exactly the same but i am sure it can be adapted to run martingale simulations.
They were using matlab and if you look around you can find some articles and even some video material about the subject.
Hi,
I think it would be quite interesting to see the outcome of Martingale- or Grid-EAs which were optimized by Atrificial Intellingence and or neural networks.
As for Martingale- or Grid-EAsthe the crash risk is very high - I would like to know what the AI or an NN would tell us.
Does anybody know a link to such an investigation? If not it might be interesting for an article?
Artificial Intelligence and Neural Networks still have to obey the laws of mathematics/physics/statistics.
So they will just as ineffective as humans at stopping Martingale strategies blowing accounts.
Artificial Intelligence and Neural Networks still have to obey the laws of mathematics/physics/statistics.
So they will just as ineffective as humans at stopping Martingale strategies blowing accounts.
This is what I expect - but a prove is something else than just an expectation. Beside that it would be a test for both sides the Martingales and the AI or NN.
AI or NN optimizing MACD-EAs is somewhat boring - isn't it?
This is what I expect - but a prove is something else than just an expectation. Beside that it would be a test for both sides the Martingales and the AI or NN.
AI or NN optimizing MACD-EAs is somewhat boring - isn't it?
Of course you can discover thresholds where the available capital vs losing streaks ratio will shift into the unrealistic domain.
And that can produce interesting results.
By all means go for it because i would certainly be interested in the outcome.
I haven't time to do this my self at the moment - :(
Therefore I asked here, whether someone has already done it (link), or might be interested in doing it (new article about different variations of Grid or Martingales tested and optimized by AI and/or NN?).
And there are some who had already developed EAs on the base of AI and NN - so it would be a lot easier for them to do this!
Sounds interesting......Go for it!
I'm also curious what kind of results you would get
I am doing exactly that! I have a Neural Network (Conv.LSTM) that predict the next high/Low price for 3 different timeframes. The Martingale strategy is not just a blind one. If there is a loss(and there are) the parameters that caused the trade for example...(difference from prediction to close price) ...are increased so as to reduce the risk of the next trade being wrong. There are several parameters that are changed if the trade goes wrong. It appears to be working well. I do limit the number of loosing trades so that it is not possible to do damage to the account. There are also quite a few other parameters that I have that come from experience..dont trade late Friday or Early Monday, for example.
I was lucky to find a good programmer to help,but unfortunetly he is a little busy with his PHD. This is a big undertaking. Just getting Meta5 to talk to the NN is quite an achievement. I started by using a CSV writer and reading the CSV into the NN but have long moved on from that.
I still need some help with making a dynamic take profit and stop loss and I am also going to change my communication setup from Zeromq as it uses a DLL and some hosting platforms dont like it.
If you are a programmer with bright ideas and especially good with MQ5, or Python I would love to hear from you.
I will try to update anyone who is interested on the progress. But so far, on this long journey, I am really happy with the results.
John
I am doing exactly that! I have a Neural Network (Conv.LSTM) that predict the next high/Low price for 3 different timeframes. The Martingale strategy is not just a blind one. If there is a loss(and there are) the parameters that caused the trade for example...(difference from prediction to close price) ...are increased so as to reduce the risk of the next trade being wrong. There are several parameters that are changed if the trade goes wrong. It appears to be working well. I do limit the number of loosing trades so that it is not possible to do damage to the account. There are also quite a few other parameters that I have that come from experience..dont trade late Friday or Early Monday, for example.
I was lucky to find a good programmer to help,but unfortunetly he is a little busy with his PHD. This is a big undertaking. Just getting Meta5 to talk to the NN is quite an achievement. I started by using a CSV writer and reading the CSV into the NN but have long moved on from that.
I still need some help with making a dynamic take profit and stop loss and I am also going to change my communication setup from Zeromq as it uses a DLL and some hosting platforms dont like it.
If you are a programmer with bright ideas and especially good with MQ5, or Python I would love to hear from you.
I will try to update anyone who is interested on the progress. But so far, on this long journey, I am really happy with the results.
John
Hi, John!
Convolution(CNN) + long short-term memory(LSTM) is great mix, especially in combination with martingale and grid trading.
I'm experience programmer and I am interested in progress in this field of machine learning, especially the practical result, but I have my own and are ready to share them too.
Hi, John!
Convolution(CNN) + long short-term memory(LSTM) is great mix, especially in combination with martingale and grid trading.
I'm experience programmer and I am interested in progress in this field of machine learning, especially the practical result, but I have my own and are ready to share them too.
yes please share
(in detail if possible)The only way to optimize Grids and Martingales is to ACCEPT THE LOSS, that is not only grid related but also to trading in general. Avoid EA that do not have loss cutoff and risks always the entire account using stop out as the only stop loss.
If you are risking the entire account you will loose it, maybe tomorrow, or next month, or next year, but it will happen.
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
You agree to website policy and terms of use
Hi,
I think it would be quite interesting to see the outcome of Martingale- or Grid-EAs which were optimized by Atrificial Intellingence and or neural networks.
As for Martingale- or Grid-EAsthe the crash risk is very high - I would like to know what the AI or an NN would tell us.
Does anybody know a link to such an investigation? If not it might be interesting for an article?