Control of the effectiveness of an Expert Advisor
Because even a live signal doesn't guarantee that any EA will keep performing the way it does in the coming future.
A live signal that is being deleted every time an EA fails or the market is going crazy, offers the same reassurance to a potential buyer as a backtest.
That's why most live signals that accompany EAs are a few weeks old and very rarely you see a long lived performance.
Also buyers need to understand that by buying an EA with a relatively affordable price, they undertake some risk about its performance and should realize that if the seller had every confidence that the EA could perform perfectly in any kind of market and had a multi month live signal to prove it, the EA's price would be in the many thousands price category and not in the 2 or 3 digit one.
My comments do not refer to EAs that are sold for some thousands, these should have a solid background that would give the potential buyer all the guarantees about performance he needs.Hi,
Why doesn't MetaQuotes require its EA sellers to post a live signal before to sell an EA on mql5 website?
A backtest is worth nothing, only a result on a live trading account has some value.
Thank you for your aswer.
In my humble opinion, I would advise against buying an EA. Always code your own, then you have full control over the quality of the output.
I do agree that back testing means nothing. Back testing is testing in markets which have reacted to different macro-economic data (i.e. war, interest rates, COVID) so exhibit different price volatilities which can stop you out more frequently than in more liquid markets. Also, it doesn't take into account the fact that you can cut your losses short before the prices hits your s/l, so your drawdown will be skewed.
There's not really such a thing as a completely hands-off EA, no two trades are the same and you can have the best EA in the world but if you have a war and a cost of living crises on the go, you need to monitor your EA much more closely and have a plan to cut your losses short. Very rarely do I actually hit my s/l, not because I always make the right can - I don't - but because I can see when the position is a "no go" and cut my losses early.
Hi,
Why doesn't MetaQuotes require its EA sellers to post a live signal before to sell an EA on mql5 website?
A backtest is worth nothing, only a result on a live trading account has some value.
Thank you for your aswer.
I agree that a backtest is worth nothing, but only if there isn't a live signal to support it. There're many ways to fake the backtesting results. Yet, once you get enough experience with EAs, you're able to spot which EA is bad and which might be good.
A live signal by itself on the other hand, is even more useless. Website which sell EAs without offering the possibility to backtest it (gladly MQL5 isn't one of them), can show you good live results of an EA and sell you another one.
The only EAs you can trust, in my opinion, are those whose live trades and backtesting trades match. Therefore yes, it's worth looking only at EAs which provide live signals.
Now, should MQL5 website force sellers to post live signals? Isn't it similar to capitalism vs. socialism topic? I mean, EAs which provide live signals are sold better anyways. It's in sellers' interest to bother about live signals.
Also, as Eleni Anna Branou mentioned, a live signal can be deleted every time the EA fails. It's up to buyers to have think critically anyways.
There are also other ways to "fake" the live signals besides "deleting it when EA fails". One would be adding money when the drawdown increases. That's what some signal providers use :(, and that's what some EA provides could do.
Another way would be to use the EA on a fishy broker. Good live results on such broker still could show bad results on more reputable brokers.
TL;DR - Some MQL5 restrictions won't replace critical thinking. There will remain many other ways to sell a terribly performing EA.
In my humble opinion, I would advise against buying an EA. Always code your own, then you have full control over the quality of the output.
Totally agree with this part. Unless you're the one who coded the EA, you can't be sure it is reliable. Even a well-performing EA on both backtesting and live, could have "adjusted" results on baktesting. One way would be simply adding a line of code such as
if (StringToTime("2020.03.19") < TimeCurrent() && TimeCurrent() < StringToTime("2020.03.26")) return;
and this way some big losses on GBPUSD could be hidden on backtesting. The casual buyer wouldn't wonder why among 1000-5000 trades, there was none open during that specific period.
You'll never have such uncertainty if it's you who coded the EA.
Now, this part
There's not really such a thing as a completely hands-off EA, no two trades are the same and you can have the best EA in the world but if you have a war and a cost of living crises on the go, you need to monitor your EA much more closely and have a plan to cut your losses short. Very rarely do I actually hit my s/l, not because I always make the right can - I don't - but because I can see when the position is a "no go" and cut my losses early.
"Cut your losses short" sounds like you're talking about grid/martingale strategies, but I might be wrong of course. In any case, many "fixed SL, fixed TP" strategies rely heavily on never closing the trade before it'd hit SL|TP. Some would call it self-discipline.
I personally prefer to not influence EA results, because as you said, backtesting "doesn't take into account the fact that you can cut your losses short". If I can't backtest such strategy, I can't rely on its statistics.
But it's a personal thing. It's like arguing about day trading vs. swing trading. Neither one is better, it's just a matter of preference.
Same here, I think all 3 options are valid: manual trading, completely automated trading, and semi-automated trading.
Because even a live signal doesn't guarantee that any EA will keep performing the way it does in the coming future.
A live signal that is being deleted every time an EA fails or the market is going crazy, offers the same reassurance to a potential buyer as a backtest.
That's why most live signals that accompany EAs are a few weeks old and very rarely you see a long lived performance.
Also buyers need to understand that by buying an EA with a relatively affordable price, they undertake some risk about its performance and should realize that if the seller had every confidence that the EA could perform perfectly in any kind of market and had a multi month live signal to prove it, the EA's price would be in the many thousands price category and not in the 2 or 3 digit one.
My comments do not refer to EAs that are sold for some thousands, these should have a solid background that would give the potential buyer all the guarantees about performance he needs.Hi Eleni.
A live signal is better and more reliable than a backtest.
Why MetaQuotes doesn't require for all EA sellers to post a 3-6 month live signal (on live account) of their EA before to sell it on mql5? This will remove all bad and not reliable EA's and buyers save money.
Even if one were to conclude that a signal is better than a back-test, it is totally impractical to force such a rule.
In order to satisfy all users, like yourself, one would need countless signals for each Currency Pair, Commodity, Index, and various other symbols, as well as for ECN accounts, for Dealing Desk accounts with large and small spreads, and all other types of variations. And for all of these, the coder would have to constantly monitor, optimise and continuously adjust, just to show that his trading skills are good.
That is not how things work. EAs are tools created by coders for traders. It is up to to the users (customers) to select and use those tools according to their own abilities, and prove to themselves that they have the skill and knowledge to wield those tools. If a user/customer does not have the necessary skill or knowledge to use an EA, then no amount of signals or back-testing is going to help them.
A trader that knows what they are doing is very capable of properly selecting an EA and using it properly to their advantage. It is only the naive, unskilled trader with little knowledge and experience, that keep wanting "proof", because they themselves are unable to achieve it, no matter what tools they use.
Hi Eleni.
A live signal is better and more reliable than a backtest.
Why MetaQuotes doesn't require for all EA sellers to post a 3-6 month live signal (on live account) of their EA before to sell it on mql5? This will remove all bad and not reliable EA's and buyers save money.
I agree that a live signal is more reliable than a backtest, but in the current financial and political environment try to run a live signal on any EA for more than a few months and tell me the results.
Even if one were to conclude that a signal is better than a back-test, it is totally impractical to force such a rule.
In order to satisfy all users, like yourself, one would need countless signals for each Currency Pair, Commodity, Index, and various other symbols, as well as for ECN accounts, for Dealing Desk accounts with large and small spreads, and all other types of variations. And for all of these, the coder would have to constantly monitor, optimise and continuously adjust, just to show that his trading skills are good.
That is not how things work. EAs are tools created by coders for traders. It is up to to the users (customers) to select and use those tools according to their own abilities, and prove to themselves that they have the skill and knowledge to wield those tools. If a user/customer does not have the necessary skill or knowledge to use an EA, then no amount of signals or back-testing is going to help them.
A trader that knows what they are doing is very capable of properly selecting an EA and using it properly to their advantage. It is only the naive, unskilled trader with little knowledge and experience, that keep wanting "proof", because they themselves are unable to achieve it, no matter what tools they use.
This is your opinion, not mine and I don't agree with your arguments. When you buy a car, this is not at the driver to make the crash test to verify that the car is not dangerous, no?
When you buy a EA from mql5, you don't have access at the EA code and you don't know how this EA is build, you follow the seller recomandation: use an ECN account, H1 timeframe, EURUSD, account of $1000 mini, etc.
8tango #: This is your opinion, not mine and I don't agree with your arguments. When you buy a car, this is not at the driver to make the crash test to verify that the car is not dangerous, no?
When you buy a EA from mql5, you don't have access at the EA code and you don't know how this EA is build, you follow the seller recomandation: use an ECN account, H1 timeframe, EURUSD, account of $1000 mini, etc.
Very good example. No matter how good a car is or how many safety tests it has passed, if a drunk or stupid driver gets behind the wheel, it is certain to crash into something or someone .
That is why, before driving a car, you have to have driver's license, to make sure you have gained the necessary knowledge and experience to be able to drive it and also to be able to test and evaluate it.
The same applies here! A user should learn how to trade properly and understand how to trade and how to use and adjust an EA, and obviously also to be able to know how to select and test an EA before buying it.
The tool (the EA), is only as good as the person wielding it.Very good example. No matter how good a car is or how many safety tests it has passed, if a drunk or stupid driver gets behind the wheel, it is certain to crash into something or someone .
That is why, before driving a car, you have to have driver's license, to make sure you have gained the necessary knowledge and experience to be able to drive it and also to be able to test and evaluate it.
The same applies here! A user should learn how to trade properly and understand how to trade and how to use and adjust an EA, and obviously also to be able to know how to select and test an EA before buying it.
The tool (the EA), is only as good as the person wielding it.But when you have driver's license and if you don't drink before to drive the car, for 90% of drivers there is no problem with the car because the crash test has been made by the constructor and driver's follow the recomandations (school driver and constructor).
With the EA, buyers follow recomandations of seller but 95% of the buyers loose money when they bought a EA because sellers sell products that have not passed the crash test (for example 6 months of live signal), there is a big problem, no?
1) This is not necesary to test an EA on demo account because you know that conditions of trading are not the sames on demo and real account and in definitively demo test it's a waste of time.
2) It is not necesary to spent time to make a backtest because it's a waste of time.
3) When you buy an EA, you have only one solution: test it on real account. But if the EA doesn't work correctly, you loose your money again (purchase of EA + capital loss).
4) Which another solution do you have to test a EA? You can have 30 years of trading experience, be the best trader on the planet, if your EA sucks you won't make any money.
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
You agree to website policy and terms of use
Hi,
Why doesn't MetaQuotes require its EA sellers to post a live signal before to sell an EA on mql5 website?
A backtest is worth nothing, only a result on a live trading account has some value.
Thank you for your aswer.