Optimize without fees

 

Hi

Is there a way to optimize an EA without any fees ? to be able to see if a strat has an edge, regardless of the fees (that skew the results)

I know you can in MT4 (just set zero spread) but I wasn't able to do it in MT5.

thanks Jeff

 
You should create a custom symbol with no swap and no commission and 0 spread fixed.

Anyway, I don't understand the sense of doing that, because all things you may found, will be not replicable in a real environment.

It's good only if your plan is to sell awesome backtests, which is a very dishonest behavior.
 
Fabio Cavalloni #:
You should create a custom symbol with no swap and no commission and 0 spread fixed.

Anyway, I don't understand the sense of doing that, because all things you may found, will be not replicable in a real environment.

It's good only if your plan is to sell awesome backtests, which is a very dishonest behavior.

thanks that's awesome i'll do that !!


when I do an optimization I want to measure the edge. For that I need to remove the fees and spread. otherwise how do you know if the timeframe is too low for example ?

you can disregard a perfectly good strategy that would work on higher TF for example. With zero spread it can be an awesome strategy but totally lose with fees, but then you know you need to raise the profit per trade. You can do that by changing the TF for example

optimizing/backtest with fees is misleading and you can skip good strategies

 
Your point of view is interesting, but personally i don't agree.

If you test your strategies in a fake environment (like what you want, or paper tests) you will have misleading results.

Every timeframe works differently, a strategy that is not profitable on M1 because of too high fee compared to expected payoff do not means that if translated to M15 become profitable.

I want all my tests to be the most reliable possible, otherwise how can I know what are the results of putting a profitable strategy studied on a fake environment on a real market? I will basically need to study it again, voiding all the time spent for the first study.

"Raise the profit per trade", seems a quick and easy thing to do, but in the reality it's a change of the strategy itself.

Anyway, the most important thing, even if I don't like the approach, is that it works for you and you can translate your strategies to real markets, otherwise you are just wasting your time, except you are a teacher or you want to sell fake hopes to someone :)
 
Fabio Cavalloni #:
You should create a custom symbol with no swap and no commission and 0 spread fixed.

Anyway, I don't understand the sense of doing that, because all things you may found, will be not replicable in a real environment.

It's good only if your plan is to sell awesome backtests, which is a very dishonest behavior.

look there should be an easier way to backtest without spreads. I dwelved into Custom Symbols 10 years ago but it was a nightmare. I'm not sure I want to get back this path again

 
Jean Francois Le Bas #:

look there should be an easier way to backtest without spreads. I dwelved into Custom Symbols 10 years ago but it was a nightmare. I'm not sure I want to get back this path again

You can simply clone an existing symbol on your broker and then change the symbol characteristics about commissions and spread.

I'm only speaking theorically, because is something I never did even if I know it should be possible.
 
Fabio Cavalloni #:
You can simply clone an existing symbol on your broker and then change the symbol characteristics about commissions and spread.

I'm only speaking theorically, because is something I never did even if I know it should be possible.

It's exactly how it should be done and it's trivial to do. Working with Custom symbol is very easy nowadays.

Though Jean-François is right, it should be possible to test with 0 spread without custom symbol. There are good reasons to do it (though Fabio, what you said about real environment is also exact). Probably MQ disabled that possibility to make it harder for scammers.

 
Jean Francois Le Bas #:

look there should be an easier way to backtest without spreads. I dwelved into Custom Symbols 10 years ago but it was a nightmare. I'm not sure I want to get back this path again

Custom symbols were introduced less than 8 years ago. So not sure what was your nightmare 10 years ago 😉

New MetaTrader 5 Platform Build 1640: Creating and testing custom symbols
New MetaTrader 5 Platform Build 1640: Creating and testing custom symbols
  • 2017.07.19
  • MetaQuotes
  • www.mql5.com
New MetaTrader 5 Platform Build 1640: Creating and testing custom symbols MetaTrader 5 platform update is to be released on July 21, 2017...
 
Alain Verleyen #:

Custom symbols were introduced less than 8 years ago. So not sure what was your nightmare 10 years ago 😉

haha!!


"You can simply clone an existing symbol on your broker and then change the symbol characteristics about commissions and spread."

no you can't

you need to program everything in detail. Zero spread is "floating"


and the programming : it's like trying to open an order in MT5 without libraries./.. could not do it. that's why I and others still prefer MT4. even if the history data on MT5 rocks. But that's pretty much it all the rest is worse (more complicated and heavy)

AND it's not compatible with MT4. big mistake


look i'll do the custom programming, but I know it will take me a long time.

Rather not

 
Jean Francois Le Bas #:

haha!!


"You can simply clone an existing symbol on your broker and then change the symbol characteristics about commissions and spread."

no you can't

you need to program everything in detail. Zero spread is "floating"

Right.


and the programming : it's like trying to open an order in MT5 without libraries./.. could not do it. that's why I and others still prefer MT4. even if the history data on MT5 rocks. But that's pretty much it all the rest is worse (more complicated and heavy)

AND it's not compatible with MT4. big mistake


look i'll do the custom programming, but I know it will take me a long time.

Rather not

You are wrong. There is no need to program anything.
 
Alain Verleyen #:
Right. You are wrong. There is no need to program anything.

I don't understand


how do I set the zero spread ? in the settings of the CS?