JAMESBR58 #: The back testing seems too good to be true. Is it possible to fool the software? I have recently bought this EA.
To the people messaging me, please note I have only used the EA for 24 hours so cannot give a full and accurate opinion. I will have further updates next week when markets open.
In the default settings, some indicators are overestimated (so that they are suitable for all the variety of situations on different currency pairs). For a specific currency pair, these indicators must be limited.
Pay attention to the following parameters: ManyManagement StopLoss TakeProfit TrailingStart TrailingStop TotalEquityLoss TotalEquityProfit LimitOrders, it is better to limit these parameters for each currency pair.
ManyManagement - the higher this indicator is, the safer the trade is, if, for example, you have one currency pair and $10,000, you can set this indicator to 10,000 and then your risks will decrease 10 times!
Assume that you need to have at least $1,000 for one currency pair. Examples of settings come from this minimum.
JAMESBR58 #: Of course, I spent $1280 just to discredit someone I do not know on the internet.
So far your bot has reduced my demo $100k to $95k in 3 days and banked zero profits.
Hi, the developers tested the data on his setup. In back testing it is working fine, but the behavior in live is always different. It is also different for me. But aparently, it is working for some people. One reason for that is that the during testing MT5 tries to simulate the movements inside bar, no matter which setup you are using.
I presented my setup to the developer and he addressed these issues. I hope it works now. But anyway this is how it works. Please contact the developer and show him what you observed.
Positions are opened according to non-network signals, with reconciliation with some indicators and with a special algorithm for dividing the position into a controlled series. These are the three main whales on which the bot is based.
For those users who want to download their own neural network, there is a corresponding free tool. With the help of which you will be able to
- conduct training in the past and verify the forecast of the trained neural network in the future.
- carry out retraining already on recent historical data, upload it to the bot and work directly with your neural network.
The bot is a tool for work, how to configure it so it will work. Of course, the bot works immediately as it is, but everyone should understand that at least you need to set up money management according to your deposit. The list of settings also includes stop-loss and take-profit settings.
Please be attentive to all details and use the bot with your head.
The bottom line is that it's hard for users to understand anything at all, and they want everything ready. Therefore, the bot stores the trained neural networks and works with them.
When you load a specific neural network, the bot will work with it, so it learns on a specific pair.
Accordingly, on the other pair you need to put either a bot with default settings, or another neural network specially trained for the corresponding pair.
This is too complicated for most users and takes quite a lot of time.
Why did you take out the Comment settings in the latest version? I had set custom comments indicating the settings that I used so I can identify which one works for my broker, now all comments are with the name and version of the bot... Can you please bring the comment back?
Mark ODonnell #: Why did you take out the Comment settings in the latest version? I had set custom comments indicating the settings that I used so I can identify which one works for my broker, now all comments are with the name and version of the bot... Can you please bring the comment back?
The bottom line is that it's hard for users to understand anything at all, and they want everything ready. Therefore, the bot stores the trained neural networks and works with them.
When you load a specific neural network, the bot will work with it, so it learns on a specific pair.
Accordingly, on the other pair you need to put either a bot with default settings, or another neural network specially trained for the corresponding pair.
This is too complicated for most users and takes quite a lot of time.
Right okay thanks, so if i was to download tickdata and train the bot over 10 years it should perform different on a specific set file with different NN loaded?
I agree it is complicated, but it appears this would be limited by data quality from the user perspective. Where is your data from tickdatasuite?
Right okay thanks, so if i was to download tickdata and train the bot over 10 years it should perform different on a specific set file with different NN loaded?
I agree it is complicated, but it appears this would be limited by data quality from the user perspective. Where is your data from tickdatasuite?
I teach the bot on graphics and in H1. If it were to teach all the techniques, I wouldn't have enough resources...
Yes, the bot will work differently for different neural network training. Despite the fact that training on the same story each time the neural network learns a little differently, especially when you configure a different structure of the neural network.
The back testing seems too good to be true. Is it possible to fool the software? I have recently bought this EA.
Assume that you need to have at least $1,000 for one currency pair. Examples of settings come from this minimum.
To train neural networks, use the utility
Of course, I spent $1280 just to discredit someone I do not know on the internet.
Explanation for everyone.
To train neural networks, use the utility
Do we need to train each pair?
Do we need to train each pair?
Of course.
The bottom line is that it's hard for users to understand anything at all, and they want everything ready. Therefore, the bot stores the trained neural networks and works with them.
When you load a specific neural network, the bot will work with it, so it learns on a specific pair.
Accordingly, on the other pair you need to put either a bot with default settings, or another neural network specially trained for the corresponding pair.
This is too complicated for most users and takes quite a lot of time.
Why did you take out the Comment settings in the latest version? I had set custom comments indicating the settings that I used so I can identify which one works for my broker, now all comments are with the name and version of the bot... Can you please bring the comment back?
So the comment field will be returned.
Wondering the same. I cant use without order comment. Please fix it asap, thank you.
Of course.
The bottom line is that it's hard for users to understand anything at all, and they want everything ready. Therefore, the bot stores the trained neural networks and works with them.
When you load a specific neural network, the bot will work with it, so it learns on a specific pair.
Accordingly, on the other pair you need to put either a bot with default settings, or another neural network specially trained for the corresponding pair.
This is too complicated for most users and takes quite a lot of time.
Right okay thanks, so if i was to download tickdata and train the bot over 10 years it should perform different on a specific set file with different NN loaded?
I agree it is complicated, but it appears this would be limited by data quality from the user perspective. Where is your data from tickdatasuite?
Right okay thanks, so if i was to download tickdata and train the bot over 10 years it should perform different on a specific set file with different NN loaded?
I agree it is complicated, but it appears this would be limited by data quality from the user perspective. Where is your data from tickdatasuite?
I teach the bot on graphics and in H1. If it were to teach all the techniques, I wouldn't have enough resources...
Yes, the bot will work differently for different neural network training. Despite the fact that training on the same story each time the neural network learns a little differently, especially when you configure a different structure of the neural network.
Demo acc looks great to be honest, still adjusting right parameters on live acc
Test table for Uni Bot version 1.7 (2 years)