Silly Qns#2 on Optimization Result

blackmore
434
blackmore  

I have these results finally after completed optimization based on Drawdown, for a period of 1 year.

My question is just so silly, but i cant help asking:

what these optimization results tells me? and what now?

also, Which pass is true, and which ones are fake?

The picture is below, and the htm file is attached for your kind considerations.


Files:
zzuegg
1503
zzuegg  

Basically nothing ;)

There are a lot of other criterias which gives you a better overview of the quality of the trades.

Search for MAE, MFE, MFEE and Risk of ruin. I cannot find the threads but there a lot of good once here on the forum. Basically with these numbers you analyze for each trade how many profit your entry/exit strategy missed during the holding time. By grouping the results on a monthly or 20Trade count or something like that and calculating the standard deviation of the results you get a good overview on the stability of the strategy. The absolute values MT4 spills out don't give you a detailed overview if you are looking for stability.

Of course as always optimize on a period of time and test on a different period to avoid falling into the overoptimization trap.

All here is my opinion and as always opinions are different. But this approach seems to work for me. Stability is the most important factor of a strategy. If a strategy is stable with a profit faktor of 1.5 it is better for me than a unstable strategy with a profit faktor of 2.


//z

blackmore
434
blackmore  
zzuegg:Of course as always optimize on a period of time and test on a different period to avoid falling into the overoptimization trap.

//z


Thats why I just opt. for 1 year, to find out 2 most important vrbs, which already took me an hour of silent agony. Hence, decided to ask you before I go on falling into this "trap".

Thanks for this. really appreciate.


blackmore
434
blackmore  
zzuegg:

Basically nothing ;)

There are a lot of other criterias which gives you a better overview of the quality of the trades.

Search for MAE, MFE, MFEE and Risk of ruin. I cannot find the threads but there a lot of good once here on the forum. Basically with these numbers you analyze for each trade how many profit your entry/exit strategy missed during the holding time. By grouping the results on a monthly or 20Trade count or something like that and calculating the standard deviation of the results you get a good overview on the stability of the strategy. The absolute values MT4 spills out don't give you a detailed overview if you are looking for stability.

Of course as always optimize on a period of time and test on a different period to avoid falling into the overoptimization trap.

All here is my opinion and as always opinions are different. But this approach seems to work for me. Stability is the most important factor of a strategy. If a strategy is stable with a profit faktor of 1.5 it is better for me than a unstable strategy with a profit faktor of 2.


//z


I read the article concerning these criterias: https://www.mql5.com/en/articles/1492

This is good, I agree.

I wish to point out, based on the mathematical treatment behind each of those, quite important assumptions are missing, some of which, this only applicable for trades or strategies that performs in sequence (cannot be concurrent trades), should have a proper time gap bet trades, (or perhaps cannot be shorter than the TF invovled), no lottage interferences, i.e fixed lots all the time. Take a deeper look into the math, and the calc.

Im also not sure why normal distribution applies? Its quite absurd - a binomial distribution maybe not so...but definitely not normal.

Ubzen
5300
Ubzen  
You cannot consider Drawdowns without considering Rewards. You'll want to choose the setting with the best risk(drawdown) to reward(profit) ratios. And don't forget to run your optimized setting on a Non-Optimized period to test stability. I like pass #25 and pass #10 is my second favorite.