Fairly compare 2 different EAs. How about 1000 different EAs? - page 7
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
There is a level of protection for Buyers in the EA Market. For example, here is one sample out of over 1000 Market EA's in my list to date (which shall remain nameless here):
10/01/2021
28
n/a
10 000 USD
So, we have an EA that's been in the Market for about 4 years, 28 Reviews were posted, all 28 Reviews were flagged resulting in zero total review stars, and its price is 10,000 USD.
I should note that this EA was far from top of Market page at the time of this post.
No real problem here, assuming that common sense is used by Buyers.
That was never be a problem, MQ used to do similar thing with the signal, if your trading style changed your previous reviews are removed. With the EAs is probably they just removed very old reviews or the ones done when the price was significantly lower than the current one. Also reviews made with fake accounts that has just 1 review and probably 1 login in the last 5 years are removed.
What I am trying to say here is that we have mainly 2 problems:
1) check the goodness of backtesting for the scam on purpose
2) check the goodness of overfitting (not overfitted)
Both are equal important but they are checked differently. While the 2 can be made by mistake or in general we can't proof the guy did it on purpose to scam people, the 1 proof without any doubts the guy is scamming the people, because if you hardcode in the EA some date to skip trades to make the backtesting look better during many years, there is no other explanation for that other than the will to scam.
For 1) I have developed few techniques that can spot these
For 2) there is even a simpler technique, that is, you download the demo of the EA on your pc, you keep it at that version for, let's say 60 days without updating it. Once 60 days passed, you backtest that version on the last 60 days that that version coudn't see, as it was downloaded before. Now if the EA is profitable or not profitable during these 60 days is completely irrelevant. It is not broker dependent as Sergey Golubev stated, becasue we do not spot it by profitability, but with the consistency between 2 periods. I will try to explain myself better. If you use the same broker to test, let's say 01/2018-03/2025 it can be profitable it can be not, maybe your broker is different(than the one used by the author of the EA), maybe has different spread, different H4 candles or timezone, or maybe you just don't have real tic data on your pc(and your history is rubbish), that doesn't matter, what matter is if you backtest the EA during 01/2018-03/2025 and then you do the backtest during 03/2025-05/2025(unseen period for the version you have downloaded 60 days before) ,using the same "environment and conditions" the 2 outcome should have the very similar metrics, very similar tendencies and very similar behaviour. If not, there is an overfit or a scam in place. And we can also use AI that do unsupervised clustering on the results to compare the 2 outcome(I did that for the sake of experimentation, becasue most classificator can estimate the distance between the 2), but believe me, most of the time you can just spot the issue just looking at the graph and comparing the stats with your brain turned to ON.
This is how I see the issue, and approach the issue, but I would gladly hear from you if you find wrong my approaches and why.
Thank you!!
It is better to inform the service desk (in case you have some proofs or you are thinking that you have it).
Because the service desk is moderating the Market.
And we may discuss some ideas on the forum as well - without concrete/technical proofs and without "too much negativity" for example.
May be ...
That was never be a problem, MQ used to do similar thing with the signal, if your trading style changed your previous reviews are removed. With the EAs is probably they just removed very old reviews or the ones done when the price was significantly lower than the current one. Also reviews made with fake accounts that has just 1 review and probably 1 login in the last 5 years are removed.
The reviews that were placed when the program was free are hidden when it becomes paid
The reviews that were placed when the program was free are hidden when it becomes paid
Thanks for that. I will add this to the footnotes in my spreadsheet.
The reviews that were placed when the program was free are hidden when it becomes paid
This is not what I have said. I have said at some point they removed the reviews made when the price was significantly lower. Of course I should specify that significantly lower is still greater than zero. But for example an EA that was 50$ got reviews and after a while it went to 3000$ but no new reviews or a few. In that case the old reviews were removed. I wasn't speaking about the case free -> paid, that was already like that in the past, at least 10 years, maybe more.
Sorry for the confusion
This is not what I have said. I have said at some point they removed the reviews made when the price was significantly lower. Of course I should specify that significantly lower is still greater than zero. But for example an EA that was 50$ got reviews and after a while it went to 3000$ but no new reviews or a few. In that case the old reviews were removed. I wasn't speaking about the case free -> paid, that was already like that in the past, at least 10 years, maybe more.
Sorry for the confusion
ow okay my bad
This is how I see the issue, and approach the issue, but I would gladly hear from you if you find wrong my approaches and why.
I would only add that the underlying cause is the fact that the overwhelming majority of traders here are OTC market traders.
If you and I were both trading Gold Futures in the centralized CME, things would be much more universal (assuming we both have up to date machines and networks).
Otherwise, it sounds like you have a good tactic going for EA selection.
This is not what I have said. I have said at some point they removed the reviews made when the price was significantly lower. Of course I should specify that significantly lower is still greater than zero. But for example an EA that was 50$ got reviews and after a while it went to 3000$ but no new reviews or a few. In that case the old reviews were removed. I wasn't speaking about the case free -> paid, that was already like that in the past, at least 10 years, maybe more.
Sorry for the confusion
Thanks. Noted for my purposes.
I think it is important to clarify something regarding your statement about MetaQuotes allegedly removing reviews.
Claiming that MQ deletes reviews, especially in a way that implies intentional manipulation, is a very serious accusation. Without solid, verifiable evidence, such claims lose credibility and can be both misleading and unfair.
MetaQuotes, like any reputable platform, has moderation policies both automatic and manual, and there are many legitimate reasons why a review may no longer appear. These include inactive accounts, deleted content, spam, violations of platform rules, and so on. Just because a review is no longer visible does not necessarily mean it was arbitrarily removed.
If you do have specific examples with dates, usernames, products, and solid evidence, I encourage you to share them in a respectful and responsible manner or contact the Service Desk to have them properly investigated.
What is not acceptable is throwing around accusations of this magnitude without clear proof. We are here to discuss things technically and constructively, not to spread mistrust based on speculation.
Thanks for your understanding.
there are many legitimate reasons why a review may no longer appear. These include inactive accounts, deleted content, spam, violations of platform rules, and so on.
Thanks. Noted for my spreadsheet again.