Forum Replies Created
-
Posts
-
The first half of the equity line before 2020 was not part of the backtest done during the generation process of the strategy. This is the reason the Balance line does not look good before 2020. When you use a start date of 2020 for the generator / reactor you do not test the generated EA before that date as there is no reason to do so as you will not be trading the EA during that time period . What you want to do is to test he strategy in the future. For example, if the generator start date is 2020 and the end date is 2022 then the EA can be tested in a demo account in the months ahead to see how it works. This is why it is recommended for students to demo trade their EAs until they gain confidence in their ability to generate EAs that will result in future profits.
To help you gain confidence the EAs created by EA Studio and FSBpro are profitable I recommend you review my demo testing of the FTMO EURUSD EA here: https://sandbox.eatradingacademy.com/forums/topic/ftmo-demo/
The forum is where students and experienced traders connect together to discuss experiences and strategies.
EA Studio does not over optimize in the generation process of creating EAs. If during the backtesting the account blows up then the lot size used in the backtest is too large causing a margin call or you have selected the option to optimize. The optimization option is not recommended to use during the generation process unless you know how to use it properly. I never use it nor does Petko use it during the generation of EAs using EA Studio. I can assure you that by using EA Studio properly it will generate EAs that can be used in live accounts. I always recommend students test the EAs they create in a demo account until they are confident the EAs created will work as expected. I recommend you download and FREE course on how to use EA Studio at: https://sandbox.eatradingacademy.com/courses/algorithmic-trading/expert-advisor-studio-free-course/
I think what you are wanting to do is to limit the risk of account loss to 1%. This is done by a calculation based upon the account size, risk in percent, and the Stop Loss. All the EAs have their individual SL set to the required number of pips to equal an account loss of 1% so you do not need to change anything. All you need to do is set the lot size based upon your account size as follows:
200K Account: Amount for new position = 6.0 / Maximum position amount = 6.0
100K Account: Amount for new position = 3.0 / Maximum position amount = 3.0
50K Account: Amount for new position = 1.5 / Maximum position amount = 1.5
25K Account: Amount for new position = 0.75 / Maximum position amount = 0.75
10K Account: Amount for new position = 0.30 / Maximum position amount = 0.30After 7 hours I decided to terminate the test. The test shows that 106,000 strategies were generated. Out of the 106,000 strategies generator 362 strategies passed the Generator filtering which also indicates 362 strategies passed the 10% OOS filtering further indicating OOS filtering is not a problem in collecting strategies. Next the 362 strategies moved to the Monte Carlo filtering were only 2 strategies passed and was moved the the Collection. This indicates that the Monte Carlo filtering heavily rejected the generated strategies indicating that to pass a large number of strategies to the Collection more time is required than the 7 hours. Personally, I was surprised that so many strategies failed to make it through the complete filtering process. I believe this could be a result of the current trading environment. It is a well known fact that the month of September is the most difficult month for trading. This could also be the reason not many strategies passed and were moved to the Collection. In order to collect more strategies it might be necessary to run the Reactor for 24 hours of 1440 minutes. Also note more than one reactor can run at the same time. A PC with an i5 processor can run as many as 3 Reactors at the same time. More than this number of Reactors simply slows down the speed of collecting strategies without realizing an increase in the number of strategies collected. A PC running an i7 processor can run as many a 5 Reactors at the same time. The test I ran today was with a PC with an i7 processor. A PC with an i5 processor most likely would not have had any strategies passing all the filtering. A solution would be to not run the strategies through the Monte Carlo process. A much better solution would be to use the ExpressGenerator to collect strategies.
Here is how it works: Lets say you are in a long position with a lot size of 6.13 lots. Now you get signal to go short. What will happen is your long position of 6.13 lots will be reduced by 6.13 lots. This then would mean that your long position lot size is now equal to zero and your long position would be closed. Now lets take another case. Let’s say you have set the reduction lot size to 1.0. Now when you get a signal to go short your long position would be reduced to 5.13 lots and you would still be in a long position instead of a short position because you still have a long position with a lot size of 5.13. Now lets take another case. Let’s say you have se the reduction lot size to 7.13 lots. Now when you get a signal to go short your long long position of 6.13 lots would be reduced by 7.13 lots. Since your long position only has 6.13 lots this would mean that your position will be changed to a short position with 1.0 lots.
After your trade makes a profit of 30 pips or more and the trade reverses direction Breakeven will place a breakeven order to protect your account from loss. The SL will not change. The breakeven order will override SL price.
I have the Reactor running right now on my machine. I have removed the count of 500 from the Acceptance Criteria. This way after strategies are collected I can go to the Collections and see what the maximum count of trades were actually collected. Give me time now to collect the data. You can also run this same test on your machine.
While the EA is loaded on the M1 time frame it also does some filtering in the M30 space. Because of the filtering done in the M30 space trades are not taken often. Sometimes it may take more than 24 to 48 hours before the first trade is taken. This then allows for higher quality of trades.
Q1 “hi how to average my loss after custom pip minus take average trade some multiplication last quantity”
A1 “Go to Account History tab at bottom of MetaTrader. Right click anywhere and save Detailed Report. At the bottom of the report you can find the average loss statistic.”
Q2 “super trend missing your list”
A2 “EA Studio only uses indicators available in MetaTrader. In FSBpro you can go to the forum and ask for a custom indicator to be built.”
I was going to suggest it might be a problem with the broker since everything we tried would not work. I forgot to add this comment in my post showing that the EA had made a trade with no errors. Not all brokers are created equal. Some brokers have large gaps in their data or even missing data. This is why it is best to use Premium data in EA Studio and FSBpro when creating strategies and this is one reason to use EA Studio or FSBpro for backtesting because you can rest assured the backtests are being done correctly.
Happy to hear we finally figured it out! Good learning experience!
Happy trading,
Alan
