ETCUSD Set File Optimum Parameters Research

Avatar
Kishan Bhatt
  • updated

Following hours of optimisation testing, losing files etc, I think I have some good research to show to try and tame the crypto beast! It has been difficult to balance drawdowns with all the turbulence in the markets this year. So these are no means perfect, but I hope they give you somewhere to start from for your aims. 


From the opt file below, I picked out the parameters 11th on the list to run as a single test. 

CommunityPower MT5.ETHUSD.M5.20200101.20200....opt

The results of which can be found below. One word of warning, these calculations were performed using the fast algorithm, so it isn't really precise. But it gives me an idea of where to try and optimise. Single tests should be done on the result you like taking into account the factors that are important to you. 

ReportTester-ETHUSD.01012020-05082020,m5.xlsx

MT version:
EA version:
Symbol:
ETCUSD
TimeFrame:
M5
Replies 9
Avatar
0
IntheDark

The ReportTester file is named ETHUSD, but inside it says Symbol=BTCUSD.

Can you share directly the .set file to compare results?

Thanks.

Avatar
0
IntheDark

My results are totaly different in Pepperstone broker with your set file.

Avatar
0
Kishan Bhatt

Hi, 

wow! What a difference!  Not sure how to explain it! I have done this with evolve markets as the broker. 

ReportTester-26546.xlsx

Not sure what the hell is going on? I notice the history quality is 99% whereas in mine its 100%. Not sure why that would be an issues though? 1% should not be enough to throw the signal off that much as it tracks it over 5 minutes in the oscillator settings. Can you send me the report of that very single test run. Below is the very settings I have used. Not to say that you have done anything wrong, but I will compare the reports side by side and go through it 1 by one. Thanks for highlighting the discrepancy, I feel it is important to check if  there are reproducible results. I hope these settings haven't gone live? I am trading with them not on a demo account to check their validity.  


Avatar
0
IntheDark

Glad to share my results.

ReportTester-107261.xlsx

Avatar
0
Kishan Bhatt

Hi! 

I am glad that you will share. I don't want my set file to be the cause of people losing money! Much prefer if we all profit. Better to work together and all that. I spent hrs trying to get a good set of parameters.

I went through it line by line, can't see any differences. But unusually, I noticed something odd about the position holding times. Your run recorded a massive length of time. Not sure if we need to run an optimisation of the filter for oscillator 1 (stochastic)? to be run specifically with this broker. In your experience, have you noticed this much variance between set files and their results with different brokers? Surely the same exchange = same ticks.

Not sure if there is something up with the broker here. But It will have the same ticks? (not sure if this assumption is valid) so not sure why there would be a difference! I will set up a demo account with pepperstone and the run the backtests to double check reproducibility. Hopefully we can tune the parameters a little. 

We can tame the Crypto!

Avatar
0
Kishan Bhatt

Hi,

If the broker also acts like the exchange in this case, won't they also set the price? I am just thinking that this will cause an issue with set files. Are you able to run the optimisation on the 1st oscillator and then another for the risk attributes? Or if you don't have time I could try?

Avatar
0
IntheDark

In the official doc file, Andrey shared the credentials of two of his broker accounts (ICMarkets and Darwinex) to do MT5 backtesting :

Doc file

With these, we can test with the same brokers to check, compare and improve results.

Avatar
0
Kishan Bhatt

Hi, 

I can't seem to get the BTCUSD symbol in those accounts. But I did a test in admiral markets account that I have.ReportTester-40938142.xlsx

This shows there is a massive difference depending on the broker you go to.