Backtesting/Optimization - page 90
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
Thanks for the reply
Can you give me an example of just such EA ie, indicator and how to check?
Thanks for the reply Can you give me an example of just such EA ie, indicator and how to check?
dasssi
The one from this post could be used as a frame for that (just set the stop loss and take profit to 0). This part :
double diffc = iMA(NULL,0,1,0,MODE_SMA,Price,BarToUse) -iMA(NULL,0,MaPeriod,0,MaMethod,MaPrice,BarToUse);
double diffp = iMA(NULL,0,1,0,MODE_SMA,Price,BarToUse+1)-iMA(NULL,0,MaPeriod,0,MaMethod,MaPrice,BarToUse+1);
if ((diffc*diffp)<0)
if (diffc>0)
doWhat = _doBuy;
else doWhat = _doSell;
if (doWhat==_doNothing) return(0);
has to be raplaced with conditions for the specific indicator, and then it could be used for optimization (so the conditions should come from one indicator and one indicator only - the one that you intend to test)
dear mladen
wich post did you refer?
dear mladen wich post did you refer?
This one : https://www.mql5.com/en/forum/176978/page9 .
Backtests with 99% modeling quality only are valuable. If your backtest is not 99% modeling quality (most common N/A) then it is absolutely inaccurate. That is very important.
Backtesting results
Hello,
I've been testing some EAs i've been coding lately and i have found out some weird behaviour on its long term performance. For example.. the last EA i tried had positive ROI since 9x until 2006-2008 and then it started to have a negative one. At first i thought this would just be a problem with the strategy itself but then after testing some other EAs i've made plus some others i downloaded (pipeater being one of them, found on this forum) i found the exact same behaviour. No matter what i do, what EAs i try, etc i always end up with shitty results after 2008-2009 period.
I wonder if anyone else is having a similar issue? at this point i'm unsure if this is a problem with changing forex rules (more false positives than before) or of i'm not backtesting correctly? (90% modeling quality btw).
Any thoughts are appreciated.
Thanks.
What's the deal with years 2008-2011 in backtesting?
I recently started developing EAs, and I encountered an issue. No matter what I try, I can't get anything working for a stable long period of time (2008-2009 are clear examples of that as shown on this image http://i.imgur.com/aBfheCB.gif).
I did some research and found that I needed a better tick data so I downloaded dukascopy data which gives me a 99% modelling quality, but the same problem still persists on every EA I create.
I've also tested some paid EAs, which had the same problem. Even found some source codes for paid EAs and saw that they were hardcoded to stop working after 2008, I guess that may be related to this issue.
So what's the deal with 2008-2011 while backtesting? Did something change in the market or what could be causing that anything I try stops working at that specific date every time?
I recently started developing EAs, and I encountered an issue. No matter what I try, I can't get anything working for a stable long period of time (2008-2009 are clear examples of that as shown on this image http://i.imgur.com/aBfheCB.gif).
I did some research and found that I needed a better tick data so I downloaded dukascopy data which gives me a 99% modelling quality, but the same problem still persists on every EA I create.
I've also tested some paid EAs, which had the same problem. Even found some source codes for paid EAs and saw that they were hardcoded to stop working after 2008, I guess that may be related to this issue.
So what's the deal with 2008-2011 while backtesting? Did something change in the market or what could be causing that anything I try stops working at that specific date every time?What is the problem with that kind of back-test results?
It looks like any normal back-testing result to me
This is what I can say about back-tests/optimization from my experience of writing an EA for almost a year and back-testing and optimizing it since.
First) It's easy to confuse back-test/optimization and curve-fitting. This is I think the mistake most people do. They curve-fit their variables for a period of time and then get bad results at another period of time.
Second) Most people back-tests their EA starting at a date in the past, let's say from 2010-01-01 to today and they are done, what about simulating forward testing by starting from 2010-01-10, 2010-01-20, and so on and optimizing as per first point 1?
Personally this is what I do, it's hard work and I obtain not the best results but the most secure ones.
Hello,
I am learning about testing EA's. Actually found this G.P Morgan-KS EA, it has martingale but i turned into 0 and works very good i think. the problem i have is that my fxcm broker only has 1 months data, would like to try doing this with 1 year but don't know where to get the history. I don't know also if it is a good configuration or what are all the variables i have changed for getting this result, it makes me about 199% profit since 6 january 2014 in GBP/USD, starting with 1000 USD lot size 0.3, tp 500 pip, sl 200 pip, 15 min timeframe. I don't know why other time frame don't work as good, but i am still happy with the result.
Where can i find more than a month data for 15 min time frame?
Where can i find a tutorial about backtesting in the forum?
How should i change the configuration of lot size, tp and sl for backtesting with other ammount of initial deposit and have same result or near this one?
Thanks
Daniel1983