Scam EAs in Market - page 3

 
Miguel Angel Vico Alba #:

We are not debating whether it is possible or not possible. Obviously there are always ways.

But understand that when I make a statement it is based on the download of a product from the Market of an average customer without knowledge that in many occasions doesn't know how an EA works, let alone how to download a *.csv, etc.

If you had a bad day don't take it out on others. Some of us are in the business of informing, not of making a point about everything that others say.

It is not a question of more or less knowledge (I don't know yours, nor do you know mine). If you think you are better than someone because you misunderstand something, maybe you have a problem.

You are the first one who attacked said that I'm talking about things that are not possible. I just prooved that it's possible and explained how. You talked about time filter (??) and DLLs, both of them really far from what I talked about.

I'm here to inform too and if you feel offended I apologize for that. I never claimed to be better than other people, it's not in my character.

About the backtestable news filter, it's the author of the EA that should provide the news database and instructions about how to perform it (if it's in its interest). For the end user it will consists only in placing the .csv database into the right place and nothing else.
 
Fabio Cavalloni #:
You are the first one who attacked said that I'm talking about things that are not possible. I just prooved that it's possible and explained how. You talked about time filter (??) and DLLs, both of them really far from what I talked about.

I'm here to inform too and if you feel offended I apologize for that. I never claimed to be better than other people, it's not in my character.

About the backtestable news filter, it's the author of the EA that should provide the news database and instructions about how to perform it (if it's in its interest). For the end user it will consists only in placing the .csv database into the right place and nothing else.

The proof of this is that 99% of Market products with news filtering are by external call and not by *csv download. Why is that? Because the developer is not interested in this analysis in the strategy tester. Only that in the future on live accounts of clients there will be no trades, which will cause a visual fantasy when the client will not get the same returns in the medium/long term. That for me is the real scam.

 
Miguel Angel Vico Alba #:

We are not debating whether it is possible or not possible. Obviously there are always ways.

But understand that when I make a statement it is based on the download of a product from the Market of an average customer without knowledge that in many occasions doesn't know how an EA works, let alone how to download a *.csv, etc.

If you had a bad day don't take it out on others. Some of us are in the business of informing, not of making a point about everything that others say.

It is not a question of more or less knowledge (I don't know yours, nor do you know mine). If you think you are better than someone because you misunderstand something, maybe you have a problem.

What he is referring to is the fact, you can integrate the data statically into the EA up to the date of release on the market.

Yes and no is because, even if you know the news, it is a problem to get real execution in Backtesting as it is in live environments. Execution in news times is very different than it is in back testing, mostly due to lack of required liquidity.

So the Backtest, even if knowing news from a static, included data array, will give different, mostly more positive results than live.

An EA can be coded to download news updates on live markets and use these data for Backtesting, filling the data gap between release and "now".

So, arguably it could be more than just a time filter.
 
Fabio Cavalloni #:
You are the first one who attacked said that I'm talking about things that are not possible. I just prooved that it's possible and explained how. You talked about time filter (??) and DLLs, both of them really far from what I talked about.

I'm here to inform too and if you feel offended I apologize for that. I never claimed to be better than other people, it's not in my character.

About the backtestable news filter, it's the author of the EA that should provide the news database and instructions about how to perform it (if it's in its interest). For the end user it will consists only in placing the .csv database into the right place and nothing else.

I am a programmer and I also originally intended to create an EA following the NeuralNetwork model.  And also realize that it is possible to generate a Backtest result based on past data, just like the author of this post said.  And I stopped how to handle data like that.  It is quite possible to create an Array of Data Structures to save instead of exporting a .csv file easily, in other words there are many ways to fake Backtest results.  The developer even built-in Array of Processed Data, creating a Fixed Array encoded in the file ex4/ .ex5 .  I also completely agree that the above practice is deceptive and immoral.  This will certainly bring negative results to the user, even their fate will become worse.  We should appreciate this.  Thank you for your good work

 
Dominik Christian Egert #:
What he is referring to is the fact, you can integrate the data statically into the EA up to the date of release on the market.

Yes and no is because, even if you know the news, it is a problem to get real execution in Backtesting as it is in live environments. Execution in news times is very different than it is in back testing, mostly due to lack of required liquidity.

So the Backtest, even if knowing news from a static, included data array, will give different, mostly more positive results than live.

An EA can be coded to download news updates on live markets and use these data for Backtesting, filling the data gap between release and "now".

So, arguably it could be more than just a time filter.

I thank you for the clarification, but it nevertheless reinforces my earlier comments about the inviavility of a "real" news filter.

Personally, I have never been a big fan of news filters for this very reason, I have always considered them to be a hoax or a fantasy. I prefer to focus on how to profit from them and not try to fight them.

Lastly, people should also be aware that there is news that is not on a macroeconomic agenda.

Edit: I still remember when the Swiss central bank came out one morning and announced that it was going to intervene in its currency. Many brokers fell that day....

 
Miguel Angel Vico Alba #:

I thank you for the clarification, but it nevertheless reinforces my earlier comments about the inviavility of a "real" news filter.

Personally, I have never been a big fan of news filters for this very reason, I have always considered them to be a hoax or a fantasy. I prefer to focus on how to profit from them and not try to fight them.

Lastly, people should also be aware that there is news that is not on a macroeconomic agenda.

Yes, I just contributed to clarify what was meant, not to counter or support any of the given arguments.

Sometimes it is necessary to filter the view of readers with additional point of views, so misunderstandings get eliminated.

I have my personal view on news and that is beyond the scope of this thread.

But in short:

 
It is fascinating how a very urgent and debatable topic is pushed into the background by a single person because he absolutely has to be right (on a completely uninteresting side issue).

Just to go back to our main topic, "Scam EAs"

Today a new EA appeared claiming to use AI and presenting a “live track record of 48 months.

If we stick to the fact that the in this case used, ChatGPT is available to the public for several months only - I'm seriously wondering how this AI magic can trade live 4 years back.

How is that possible, if we consider that the used AI wasn't available at the time this EA started?

So my question is, is it possible that “cooperative” brokers set up a “live account” that is purely based on backtest data?

I'm just thinking loud... 
 
Daniel Stein #:
It is fascinating how a very urgent and debatable topic is pushed into the background by a single person because he absolutely has to be right (on a completely uninteresting side issue).

Just to go back to our main topic, "Scam EAs"

Today a new EA appeared claiming to use AI and presenting a “live track record of 48 months.

If we stick to the fact that the in this case used, ChatGPT is available to the public for several months only - I'm seriously wondering how this AI magic can trade live 4 years back.

How is that possible, if we consider that the used AI wasn't available at the time this EA started?

So my question is, is it possible that “cooperative” brokers set up a “live account” that is purely based on backtest data?

I'm just thinking loud... 

without a doubt, there are many dodgy brokers out there, but we are not allowed to discuss them, meh..

 

I'm gonna post some counter arguments here :

If i over optimize an ea with a genetic algorithm and i publish it , what is the difference of optimizing the ea with a neural network (i mean actually optimizing the neural network in the ea) . I can overfit on both cases . The tester is a genetic algorithm . So , should optimized eas be banned because people may be tricked ?

I believe that among the authors that deliberately found the "trick".... 

The trick :

  • History reader
  • Go to fiverr , print screen a logo template , and whatever is on that logo is now the name of the ea [if you wanna have a good laugh and you are out of comedies go to a product copy the url of its logo , go to google lens and do a reverse search for that image , get pop-corn and then think what does it say for the vendor that he stole the sample from a collection of graphic designs]
  • Sprinkle keywords , AI , gpt etc
  • Post screenshots of millions , videos too 
  • Make promises of $ even tho is not allowed
  • Publish
  • Create FOMO because they unfortunately only have 10 copies left
  • If a bad review comes complain 
  • solicit reviews to get "proof of purchase" for an exclusive group or something else

...there are authors that accidentally are sitting on a slight overfit which works for a while and are unaware of it.

So there's vendors who actually think they found the holy grail .

Now here is the difference between those people and scammers . The ones above that have some wind in their sails ,when they realize that it was not the holy grail they may actually care about their customers and increase their efforts try to learn more things , and with a bigger budget they can now hire someone else to guide them around some concepts.

The news filters values are dead wrong sometimes , but most are using it for avoiding high risk events so ... 

Edit : 

Forum on trading, automated trading systems and testing trading strategies

Scam EAs in Market

Dmitry Zhakov, 2023.07.25 10:34

I recommend to MetaQuotes to hold the money after the client purchases any product for at least 3 months. And if no complains from client, then release the money to the seller. Otherwise to refund the money to the client. This will push away scam EA sellers.

In this time, the client can test the product and see of there is something wrong. 


Also with all due respect , 3 months is absurd 
 
Daniel Stein #:
It is fascinating how a very urgent and debatable topic is pushed into the background by a single person because he absolutely has to be right (on a completely uninteresting side issue).

Just to go back to our main topic, "Scam EAs"

Today a new EA appeared claiming to use AI and presenting a “live track record of 48 months.

If we stick to the fact that the in this case used, ChatGPT is available to the public for several months only - I'm seriously wondering how this AI magic can trade live 4 years back.

How is that possible, if we consider that the used AI wasn't available at the time this EA started?

So my question is, is it possible that “cooperative” brokers set up a “live account” that is purely based on backtest data?

I'm just thinking loud... 

Why do you interpret that the use of AI in an EA is obligatorily linked to ChatGPT?

Beyond ChatGPT, AI has been used for several decades to a greater or lesser extent.

We have been using neural networks, deep learning, etc. for decades, and they are all AIs.