Backtesting/Optimization
During backtesting we may have the several cases:
- For example, some EA is testing very good, perfectly well: it does not mean anything for me because code of EA may be adapted by programmer to be tested perfectly well.
- if the EA is showing very bad results during the backtesting I will look on the original idea trying to improve something in original idea.
- if the EA is testing but sometimes good and sometimes bad (just for example: good testing during the October data, and bad during the September, good for August etc) this EA is very interesting for me. Because I understand that it is impossible to have the stable good results forever (because market is changing and everything is changing but we are using the same indicators and the same EAs and is not changing anything).
i think the strategy tester is a good filter, it does show if a strategy has promise, reveals strengths and weaknesses of a given EA. Optimization helps mitigate the weaknesses and exploit the strengths. Live demo testing ensures that given real time live prices, the EA efficiently communicates with the Broker's server & executes as intended. Live real money testing proves the EA, with real results. whether it is profitable or not.
with mt4, one is able to "live real money" test with lot sizes as small as $100, or a penny a pip. given that brokers such as IBFX pay interest on demo accounts, but not on live mini accounts, i believe it is important to "live real money test" with the smallest lot size allowable to be sure that the EA can overcome all obstacles presented to it, eg, swap charges, mid day build upgrades, isp interuptions, nfp days, etc, etc...
my humble opinion only... 
hey .. my opimion with ea is well they are ok but they do not tell what will happen in future price only past history so 1 expert might work well 1 or 2 years then might work as it did
Jannik
I have a book in danish and the most common strategy is as simpel as a moving average above below but here u miss some of the top and bottom
my humble opinion only...
additionally, see the links listed here
http://www.strategybuilderfx.com/forums/showthread.php?t=15309
http://www.strategybuilderfx.com/forums/showthread.php?t=15253
хорошее везение, хороший торговать!
Who knows if MT4 strategy tester takes into account highest and lowest prices of the bar or just open/close prices ?
Maybe new strategy tester built form scratches needed for testing EA's..
That way we could have complete control over it..do we ?
MT and backtesting
hi all,
i am new to this forum and would like to start with some questions regarding backtesting im MT.
i read on the net, that the backtest results of MT cannot be relied on.
can anyone really confirm this?
is there a serious bug in MT?
i can imagine, that the reason for this is in most cases only bad system programming.
how about the bar handling in MT?
lets say we look at daily bars.
does strategy tester only look at OHLC?
or does it look at every single tick internally?
this fact is important to know.
behaviour will differ in these 2 scenarios, if we have 2 or more signals on the same daily bar.
thanks.
thanks for switching me here.
all questions answered in fxid10t's links.
many thanks.
Backtesting
Hi, everybody.
I'm new to this forum and english is not my native language.
First of all I would to congratulate you for the high quality of the posts. Not common in other forums I have visited.
I'm playing forex and coding EA for some months.
My main trouble is why I get so higher profits (even +1000%/month) when backtesting my EA than on live? I have tried many different strategies and none result noticeable while terrific at backtest.
Support says that backtester uses just OHLC data but that is not true as I see price changing within the bar on strategy tester. By the way, I use Metatrader 3.83 build 6231 from InterbankFX.
Can anybody be of help?
Thanks in advance
Renato
From my experience strategy tester of MT3 has at least one serious bug.
That bug hits you, if you are using limit orders (OP_SELLSTOP, OP_BUYSTOP).
For example, if your buy limit is over the actual price, then strategy tester will execute this order with the actual price.
In real life this trade would not have taken place, because buy limit was not reached.
Like I said, this is from my exprience.
Maybe some others can confirm this?
To be sure you could transfer your EA to MT4 and test ist there.
Results should be more realistic.
Yes, Iomme. I agree with you.
I've have the feeling of something like that happening. I noticed stop orders being processed quite easier than in live mode.
But I fell it's not the only problem with ST. I suspect even normal orders are fullfiled somewhat easier like there were no slippage or something like that.
The EA's I have written scalp to profit in ST making huge profits but get frequently 'invalid prices' or go stoploss in live mode.
I would love to understand how exactly ST inflate simulated profits and be able to adapt EA to get part of that good profits.
Renato
From my experience strategy tester of MT3 has at least one serious bug.
That bug hits you, if you are using limit orders (OP_SELLSTOP, OP_BUYSTOP).
For example, if your buy limit is over the actual price, then strategy tester will execute this order with the actual price.
In real life this trade would not have taken place, because buy limit was not reached.
Like I said, this is from my exprience.
Maybe some others can confirm this?
To be sure you could transfer your EA to MT4 and test ist there.
Results should be more realistic.
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
You agree to website policy and terms of use
It was a problem which renoex arised https://www.mql5.com/en/forum/general
Not asked even. It was just 'rhetorical' questions (question with answer inside the question). Of course no we can not.
But we are working with this tester and do not have any choice.
1. Some people say: do not believe in mt4 strategy tester. To understand about the particular EA you need to test it on demo during the several years (5 or 8 years).
2. The other people (programmers) say that do not believe in demo testing as well. You need to use real money (during the 5 or 8 years) to say: this EA which I (programmer) created are good (or bad).
In this case we have the following: programmers proposed some EAs, testers are spending their own money to prove the programmer's work. And nobody is responsible for anything of course.
3. The other people say that it is not enough even. Because we need to test on real money using the different brokers and different timeframes as well (but nobody said where to get this money from ...).
3. Some people are using mt4 strategy tester to say something about particular EA.
What is your choice?
How the people testing?