Newdigital's EAs

To add comments, please log in or register
dvarrin
279
dvarrin  

Hi new digital,

I've seen you have implemented lots of EAs and I wanted to ask you if you could tell us which ones you like, which ones are giving the best results, which ones are more secure, ... Only a small summary. I don't ask for all the statements ;-)

Actually I'm testing Predator. It's a really great one. Have you seen other systems near so good like Predator?

Cheers,

Daniel

Sergey Golubev
Moderator
116927
Sergey Golubev  
dvarrin:
Hi new digital,

I've seen you have implemented lots of EAs and I wanted to ask you if you could tell us which ones you like, which ones are giving the best results, which ones are more secure, ... Only a small summary. I don't ask for all the statements ;-)

Actually I'm testing Predator. It's a really great one. Have you seen other systems near so good like Predator?

Cheers,

Daniel

Hi dvarrin,

It is very easy to see: just download excel files from the first post of sticky files' thread. But please note: not every EAs performance is mentioned in excel files. So you should also look at total leaders, weekly leaders and weekly performance thread.

dvarrin
279
dvarrin  

Ok, I'll check. I was only thinking that there could be some great EA which are not tested. Like for example Predator :-))

Sergey Golubev
Moderator
116927
Sergey Golubev  
dvarrin:
Ok, I'll check. I was only thinking that there could be some great EA which are not tested. Like for example Predator :-))

Yes, you are right: there are a lot of EAs which were not tested yet.

https://www.mql5.com/en/forum/176044

Sergey Golubev
Moderator
116927
Sergey Golubev  

Besides, some EAs are still under development and improvement. Anyway, we have enough new/untested EAs to be tested for the whole next year for example.

Anyway, all EAs (bad ones, good ones and anyones) are listed in sticky files' thread.

dvarrin
279
dvarrin  

Hi newdigital,

I was testing some EAs and I could see that for some of them, the tests were quite good when using the open price, but using the ticks it became really bad for some.

Have you or anybody ever tried to update the EA so that it enters only on closing or opening of a new bar?

Cheers,

Daniel

Sergey Golubev
Moderator
116927
Sergey Golubev  

All the EAs were coded on close bar except the following:

- Firebird was coded to trade on close bar, open bar and high/low of the close bar. This EA is having 3 modes of trading switching between each other automatically depends on the day of the week and depends on profit/loss of the previous trade. It was coded like that.

- one of indicator inside fozzy EA is acting on open bar but it is because of my settings.

- MAChannel was coded on high/low of the close bar. Close bar but high low (high/low of the close bar is very different for different broker).

Any other EAs are testing and coded on close bar only (previous close bar if we consider current bar as open only).

dvarrin
279
dvarrin  

Hi,

Thanks a lot for your answer. So I guess that using the open price for optimisation is the best some some EA like MAChannel? I am trying it and yes it gave really good results using the opening price, but when I went to tick price it was not so good at all.

I got the historical data from Alpari to get 99% modelling accuracy. But what will happen if I optimize a system using those data? I'm using IBFX for my demo account and I would like to open a live one. If IBFX is giving different values for the candlesticks, if a system is giving good result with IBFX data, why is it not considered to be a good modelling?

Cheers,

Daniel

Sergey Golubev
Moderator
116927
Sergey Golubev  

MaChannel, as I remember, was coded on close bar. But high/low of the close bar. I have no idea why backtesting is not the same with forward testing with this EA.

I backtested some EAs in portfolio section here https://www.mql5.com/en/forum

If you are using IBFx so you should optimize with IBFX only. Because data for Alpari rus and IBFX is very very different.

Or just attach MaChannel with IBFX demo and compare IBFX trading with my North Finance trading.

dvarrin
279
dvarrin  

Hi new digital,

Thank you very much for you answer. So if I understand well, the people who are using the alpari data pretending to have a 99% accurate modelling are a bit wrong?

I'm sorry to ask you so many questions recently, but I'm really happy that you have time to answer and always your answers are really helpful

About the martingale EAs. Is there an EA which takes into account the volatility and uses a higher number of pips when the market is trending a lot?

Regards,

Daniel

Sergey Golubev
Moderator
116927
Sergey Golubev  
dvarrin:
Hi new digital,

Thank you very much for you answer. So if I understand well, the people who are using the alpari data pretending to have a 99% accurate modelling are a bit wrong?

I'm sorry to ask you so many questions recently, but I'm really happy that you have time to answer and always your answers are really helpful

About the martingale EAs. Is there an EA which takes into account the volatility and uses a higher number of pips when the market is trending a lot?

Regards,

Daniel

Alpari data is good. Maximum may be 90% with metatrader's strategy tester with any data. 90% is fine. Alpari data is similar with North Finance data, with Fibo Group data and sometimes with LiteForex data. And Alpari data is different from FXDD and IBFX and if people are using EAs with my preset with FXDD or IBFX so it is the bettrer to optimize the settings with those data as well just for some period of time to correct the setting.

Some people may backtest EA and compare backtesting with forward testing. And they may find differences. Why?

Because if EA was coded on open bar, or using open bar or close 0 bar, or high/low of the bar and so on so backtesting is not reliable. And if some person is selling EA based on backtesting results and we can not look inside the code to see this "open bar, or using open bar or close 0 bar, or high/low of the bar and so on" so we can not belive in backtesting in this case.

The other issue is the following: some indicators are having different value (showing differently) with different brokers. And if EA was coded using those indicators so we can not rely on backtesting.

For example: Brainwashing EA. This EA is using iTrend indicator. iTrend indicator is related to concrete broker's data only: i may attach indicator to Alpari chart and will get 0.0025 (value of the indicator). You may attach indicator to IBFX chart and will get 0.0028. Different data - different value of iTrend indicator and different results for Brainwashing EAs during the forward testing or backtesting. That is why we are calculating the value of this indicator manually every week for this EA (i am doing it every week and i call it 'EAs with dynamic iTrend value').

Thus, Alpari rus data is fine for backtesting and 90% is fully enough. But to speak about backtesting for concrete EA we should know some issue about this EA, we must see the code of EA to be sure that we may rely on backtesting. Alpari data + original code of EA will give us some understanding of that.

Unfortunately, IBFX and FXDD brokers are not providing the data for backtesting for 2 or 4 years.

----------------------------------

Commercial EAs.

Just for example: some seller is selling EA to you in commercial section. Seller is providing backtesting (good one).

- Can we see inside the code of his EA to know how this EA was coded ("open bar, or using open bar or close 0 bar, or high/low of the bar and so on")? No, we can not. Because he is selling ...

- Do we know indicators which were coded inside commercial EAs ("some indicators are having different value..."). No. Because his EA is commercial one.

- If the seller created the settings using Alpari rus broker's data so can we use same setings for IBFX? No, we can not. We should optimize this EA and create our own settings.

- Can we optimize the settings of commercial EA with expiration date? No, we can not.

- Does the seller know everything which I post here? Yes, of course.

That is why it is not good to sell EA based on backtesting results only.

--------------------

As to martingale EAs so I don't know.

Your question about it is more related to money management (MM).

There is some library file posted somewhere (check files' thread or links' thread) with more than 20 MM functions inserted in any EA autiomatically so it is more related to MM.

To add comments, please log in or register