For a sustainable EA, how much win% should be needed at least?

 
  • 13% (8)
  • 12% (7)
  • 8% (5)
  • 20% (12)
  • 17% (10)
  • 10% (6)
  • 20% (12)
Total voters: 41
 

Dear experienced traders, 


How do you think for a sustainable EA, how much win% should be needed at least?

Other points such as profit is more important than win%,  but win% is still an important question to design strategies. 

Over optimization is horrible, even that we have to keep win% at least. 


If you are using many kinds of strategies, you can choose more than 2 choice. 


Thank you in advance and best regards

MQL5 forum
MQL5 forum
  • www.mql5.com
MQL5: Forum on automated trading systems and strategy testing
 

What is the point of this poll?

There is no right answer as it all depends on the Risk:Reward ratio.

 
Keith Watford:

What is the point of this poll?

There is no right answer as it all depends on the Risk:Reward ratio.

Thank you, Keith. 

I'm having some study about some strategies, some of them are very special. 

I'm not sure there is a right answer or there is not a right answer, but we have to find a reasonable answer... 

 

A classic mistake is to focus solely on win.

You can also try to focus on limiting loss.

For example, when i set my max loss for my next trade to -$100 then i will not know any win yet, but i do know that i will not lose more then $100 on my next trade.

So i have successfully 'predicted the future' in terms of the worst case scenario without ever focusing on win.

 
Usually EA opens a lot of trades.  In fact, too many.  I would surmise a high win rate is needed to be profitable.  Of course R:R matters more, but a great R:R ratio is not always possible in reality. 
 

The last Basel implementations that I have seen had 5 statistical groups/levels of risk

1 group => loss from   0% to   5%  <= acceptable     the best

2 group => loss from   5% to  20%  <= acceptable

3 group => loss from  20% to  50%  <= not acceptable but continued

4 group => loss from  50% to 100%  <= not acceptable need to terminate

5 group => loss    more than 100%  <= not acceptable extremely bad
This is the reason I voted for more than 80% win


 
AIRAT SAFIN:

The last Basel implementations that I have seen had 5 statistical groups/levels of risk

1 group => loss from   0% to   5%  <= acceptable     the best

2 group => loss from   5% to  20%  <= acceptable

3 group => loss from  20% to  50%  <= not acceptable but continued

4 group => loss from  50% to 100%  <= not acceptable need to terminate

5 group => loss    more than 100%  <= not acceptable extremely bad
This is the reason I voted for more than 80% win


Thank you, Airat. 

80% looks very over optimized. 

May I ask how long will be your back test for? 

For example: 1 year, 5 year...

For a sustainable EA, how long period back test should be needed at least? (with 99% high quality back test data)  
For a sustainable EA, how long period back test should be needed at least? (with 99% high quality back test data)
For a sustainable EA, how long period back test should be needed at least? (with 99% high quality back test data)
  • 2020.12.28
  • www.mql5.com
Less than 6 months will be fine, More than 6 months, More than 1 year, More than 2 years, More than 3 years, More than 5 years, More than 7 years...
 
Marco vd Heijden:

A classic mistake is to focus solely on win.

You can also try to focus on limiting loss.

For example, when i set my max loss for my next trade to -$100 then i will not know any win yet, but i do know that i will not lose more then $100 on my next trade.

So i have successfully 'predicted the future' in terms of the worst case scenario without ever focusing on win.

Thank you so much, Marco. 

I got it. 

 
Koh Kok Yeow:
Usually EA opens a lot of trades.  In fact, too many.  I would surmise a high win rate is needed to be profitable.  Of course R:R matters more, but a great R:R ratio is not always possible in reality. 

Thank you, Koh. 

Yes, it depends on the strategies, scalp have to win a lot and trend do not need the have a high win%. 

 

Sky L:

May I ask how long will be your back test for? 

For example: 1 year, 5 year...

In most cases 2 weeks