90% Backtest Data

[Marco vd Heijden]

Neglectable 

[Waseem Raza]

Modeling Quality does not means EA quality  it means broker graph M1 history graph quality 

[Ovidiu Caslariu]

Probably enough to try it for real.

[Erdem]

tick simulation is not realistic enough, so if u do it via tick simulation it doesnt represent the work on real conditions. and when u use 1M data it can always misrepresent the entry and exit points. so you have to try the EA on real time data or on real tick history.

btw the new feature of MT5, seems enough to solve this (real tick history data use on tester), so in the near future we can trust backtests about that it does really represent the real conditions.

on the other hand there is a serious question: is the real conditions of market, stable enough to represent the future conditions?

that is the issue, and always will be..

so with the current features of MetaTrader i think it is (%90) not good enough. 

[Marco vd Heijden]

The real question here is:

90% Backtest data,

Good enough, For what ?

[Alain Verleyen]

Marco vd Heijden:

The real question here is:

90% Backtest data,

Good enough, For what ?

Marco, you should create a topic to explain your view about the backtester.

It will be more constructive and positive that always questioning people.

[Osama Shaban]

I use back-testing in only to get a general idea about will I consider there is a hope in this EA ? But it is not final. Forward testing is essential.

Also use it to check the visual mode to know if my EA works exactly as I must to do or to reveal others EAs ideas. 

So, even if I got 90% modelling quality, it is not enough to use it on live account. 

[Marco vd Heijden]

Alain Verleyen:

Marco, you should create a topic to explain your view about the backtester.

It will be more constructive and positive that always questioning people.

I learned over the years, that if someone is on collision course, nothing can or will stop them from achieving total destruction.

[Lorentzos Roussos]

Erdem Sen:

tick simulation is not realistic enough, so if u do it via tick simulation it doesnt represent the work on real conditions. and when u use 1M data it can always misrepresent the entry and exit points. so you have to try the EA on real time data or on real tick history.

btw the new feature of MT5, seems enough to solve this (real tick history data use on tester), so in the near future we can trust backtests about that it does really represent the real conditions.

on the other hand there is a serious question: is the real conditions of market, stable enough to represent the future conditions?

that is the issue, and always will be..

so with the current features of MetaTrader i think it is (%90) not good enough. 

I agree . 
Thats is a major plus on shifting to MT5.

What if -even though unorthodox- we modify our EA to trade with the same conditions the strategy tester offers instead of 
modifying the way we conduct backtests.

For example , we know for sure Time , Open and Close are Correct and pretty impossible to mess up during simulation.
What if we enter on Opens only (this is the case for most strategies nowadays anyway) ,but also check our trailing on Open
and manage stops ourselves on Open too ?

The biggest issue - and the only issue - the MT4 Strategy Tester has is you dont know what was reached first SL or TP,
and by managing everything ourselves we run into that issue again . Cant avoid it .
It comes down to what are you checking for first SL or TP ?

[Rogerio Figurelli]

Lorentzos Roussos:

The biggest issue - and the only issue - the MT4 Strategy Tester has is you dont know what was reached first SL or TP,
and by managing everything ourselves we run into that issue again . Cant avoid it .
It comes down to what are you checking for first SL or TP ?

Right, SL/TP are real time, but as you stated before, you can manage such problem in the EA too, using not so small SL/TP values.