A balance between QUALITY and QUANTITY

 

I am interested in the following question: how to choose the best balance between quality and quantity when optimising an EA.

Here are some examples to make it clear what we are talking about:

1. The highest profit. However, the number of trades is not very high. It has a very large drawdown %:

2. Highest number of trades. However, low profit and high drawdown %:

3. Highest expected payoff. However, number of trades is extremely low, but drawdown is OK.

etc. ....

I am interested in a mathematical criterion that allows estimating the best ratio of trades quality and profit.

Another question concerning testing of Expert Advisors. Does the tester have any limitations on the number of optimized parameters and optimization cycles? I noticed that the tester simply rejects to conduct testing if there are too many cycles. How to fix this problem?

 
Personally, I first filter by profit, as this is basically the only thing I came to forex for, then I watch the drawdown and look for the optimum profit/ drawdown ratio
 
In short, in this case I would choose the first option
 
RomanS: Personally, I first filter by profit, because this is basically the only thing I came to forex for, then I look for the drawdown and look for the optimal profit/ drawdown ratio

I.e. for you: Criterion=Profit/Percentage of Drawdown Am I right?

 

There is no single criterion, as I'm sure you know, Sergey. Every trader develops his or her own, which takes into account the parameters that he or she considers most important. Perhaps this thread will offer you a criterion that is to your liking.

Of the three passes posted, the last one, although with minimal drawdown, is statistically unrepresentative. There is simply nothing to talk about. The other two... You can see for yourself what the drawdowns are.

As for the tester - it would be better to read the articles posted here. There are probably limitations, but with a reasonable approach to optimization, you can consider that they simply do not exist (they are much higher than reasonable limits).

P.S. Frankly speaking, it's a long time since I dabbled in optimization. Or the tester itself :)

 

I am not looking at % but at $ because % depends on the initial deposit, if you increase it by 10 times, the % of the drawdown will decrease by the same amount.

 
Mathemat:

There is no single criterion, you probably know that, Sergei.Every trader draws up his own, which would take into account the parameters he considers most important .


With my 1000th post, I want to agree with you 100%
 

In my opinion, there are other important parameters by which an EA should also be evaluated.

For example, the proportion of time the Expert Advisor works with the deposit, if I can put it that way. If the share is low and the money is idle most of the time, then most likely, we can add something else to the Expert Advisor to involve the deposit during this idle time and increase the profit of the Expert Advisor.

I will ask one more question. What is the maximum drawdown percentage that is acceptable at constant lot? I personally think it is 20%. I understand that there are no clear criteria, but my thoughts on this subject are interesting.

 
Richie:

In my opinion, there are other important parameters by which an EA should also be evaluated.

For example, the share of time the Expert Advisor works with the deposit, if I can put it that way. If the share is low and the money is not working most of the time, then most likely, we can add something else to the Expert Advisor to involve the deposit during this nonworking time and increase the profit of the Expert Advisor.

I will ask one more question. What is the maximum drawdown percentage that is acceptable at constant lot? I personally think it is 20%. I understand that there are no clear criteria, but my thoughts on this subject interest me.

The main thing is to decrease losses and increase profits.

System (TP-170, SL-10, drawdown - 6%, profitability - 3.12, expected payoff - 2400). Run in 10 years.

 

I, too, have been thinking about this for a long, long time. I will offer you my criterion, but unfortunately I cannot see it in the tester. If anyone likes it and can help implement it, it will be very interesting.

1. The calculation is only in points, no money.

2. The main criterion when you enter the trade, the price should not go against you. The best system is the one with the minimum drawdown in pips (during the time of existence of transaction ). The ideal TS drawdown is 0.

3. No averaging and lots. There may be only 1 order in the market for one instrument.

4. After tests. Selection of the 5 best systems according to criterion 2 (statistically and maximally (minimorum)). And check it on the forward section.

5. ....

I forgot to add SL and TP no (when searching for the best TS). when trading on the real SL is MUST!!! (3*sigma of the drawdown obtained in the tester)
 
Mathemat:

There is no single criterion...

what is such a criterion in your opinion ?