Download MetaTrader 5

EA test results - what is real and what is not

To add comments, please log in or register
Have you developed a profitable strategy? Launch your trading signal and make profit from it!
Daniel Petrovai
440
Daniel Petrovai 2014.04.27 01:40 

Hi guys,

I wanted to post a couple of pictures, one for the report and one for the graph of an EA that allegedly can return spectacular profits.

The rule is quite obvious, if it's too good to be true, it usually isn't true, but I wanted to hear more opinions about the theoretical possibility to achieve such results.

So, what do you think, is it possible, in theory to have this working or not ?

The graph (note the 66 million figure)

and the Report

Presumably, the test was done on : account star at $ 10000 dollars,1:400 margin and test period 10 - Jan 2011 to 25 - Apr 2014 :

and last trade

I found the last 2 trades a bit of mind blowing since on closer look order 70556 hit SL on 2014.04.25, then Sell 70557 was open on 2014.04.25 and hit the stop after 10 pip but the date on it is 2013.03.24

Then the next ticket 70558 was open in 2013.03.24 and close at stop on 2013.01.28 ?!? for a profit of 1.7 pips

Is that a tester bug or something else ?

The price looks real and quite there, at least what can be seen from this pics.

So, how about it ?

Is it theoretically possible to achieve such fantastic results or it's a fake ?

SYED NAUFAL GADDAFI
1401
SYED NAUFAL GADDAFI 2014.04.27 01:54  

First of all, you need to check the tick history and check if the date matches.

Secondly, in real world, who ever makes money, broker will start to intervene and all sort of problems will start to arise.

Delayed executions, unreal slippage, unable to close winning trades and so on.

Backtesting is a near perfect world. That is why crappy EA can still makes money at some point and then blows up.

Ian Venner
2584
Ian Venner 2014.04.27 02:25  

Backtesting results can be faked in lots of ways, the only way to find out is by doing your own testing. My advice is do not buy an EA without a considerable free trial period. If the developer knows his EA is good he would be happy to allow you a testing period.

Daniel Petrovai
440
Daniel Petrovai 2014.04.27 02:40  
SDC:

Backtesting results can be faked in lots of ways, the only way to find out is by doing your own testing. My advice is do not buy an EA without a considerable free trial period. If the developer knows his EA is good he would be happy to allow you a testing period.


That's true and good, sound advice.

I was asking if in theory it is possible since by my calculations, for a 1:400 account, it would require a merely 500$ a trade profit to reach the 164 million at the end of the period.

Considering that the lots are growing with the account, it would imply quite small profits to start with and grow with the lot size.

Ie : if auto lots used, 10000 equity start and lot size = MINLOT x equity/1000 as I found somewhere on the forum, it would start with 0.1 lot and get to MAXLOT when required equity is available.

I understand there's more involved than a good EA to reach such results, that's why I was curious if such results are theoretically possible.

Possible some of you guys are more up to date with the results achieved in trading competitions and things like that. Is this sort of results they get there?

luxinterior
385
luxinterior 2014.04.27 22:32  

thrdel the 'theory' of anything is possible but at the end of the day that's all it is; 'theory'.

In addition to the sound advice you've been given above, another thing to point out is the modelling quality of that back test is only 25% so it's definitely not to be relied upon.

Hope that helps and good luck!

Lux

Daniel Petrovai
440
Daniel Petrovai 2014.04.28 02:13  
luxinterior:

thrdel the 'theory' of anything is possible but at the end of the day that's all it is; 'theory'.

In addition to the sound advice you've been given above, another thing to point out is the modelling quality of that back test is only 25% so it's definitely not to be relied upon.

Hope that helps and good luck!

Lux


Thanks for your reply luxinterior:,

I take it that your answer is yes to " is this possible in theory ?" question, right?

I'm not looking for advise to buy this thing or anything, the plan is to hack the hell out of it, expose it's guts, still the code and use it for free .

Just kidding you, it's a project I'm working on that returned stupidly good results like that.

SYED NAUFAL GADDAFI
1401
SYED NAUFAL GADDAFI 2014.04.28 02:33  
Well, hopefully can see a forward test. :)
Ian Venner
2584
Ian Venner 2014.04.28 02:47  

Yes and if it makes anything like $164 million dollars tell the developer we want one too ;)

MisterDog
381
MisterDog 2014.04.28 03:02  
I'm curious, is your program highly optimized? I believe these results may not be so amazing if the EA were tweaked to fit the historical data. If your program is not optimized then I am impressed.
Daniel Petrovai
440
Daniel Petrovai 2014.04.28 03:52  
deysmacro:
Well, hopefully can see a forward test. :)

I have my eye on it, believe me !
Daniel Petrovai
440
Daniel Petrovai 2014.04.28 04:08  
MisterDog:
I'm curious, is your program highly optimized? I believe these results may not be so amazing if the EA were tweaked to fit the historical data. If your program is not optimized then I am impressed.


To satisfy your curiosity, yes, the program was optimized but no, it wasn't highly optimized. It was optimized on this year data - approx. 4 month - and then used the same settings for 1, 2, 3 year period.

I'm also curious to see how it performs in real market conditions. Only time will tell .

12
To add comments, please log in or register