Which EA has the best win/loss ratio?

New comment
 

For example....

in this EA

https://www.mql5.com/en/forum/trading_systems

the third position in the series of five has the following win/loss report

winners 1273

losers 184

therefore it's win/loss ratio is

the number of winners/total trades or...

1273/1457=0.873713109=87%

can any other EA beat that I wonder?

To figure this out I isolated each position the EA took based on it's lot size. I could do that because this EA doubles the lot size for each trade in the series of trades it executes when it trades. I copied the strategy test results into excel and simply sorted for each trade of the same lot size and then counted how many won and how many lost for each of the five (when maxtrades=5) that it executes.

87% is pretty good I think. but I wonder if there are any better?

Anyone know of a better win/loss ratio?

 

i have been using a more profit and loss system of measurements. i am sure that using both systems together would net some interesting results.

 
Aaragorn:
For example....

in this EA

https://www.mql5.com/en/forum/trading_systems

the third position in the series of five has the following win/loss report

winners 1273

losers 184

therefore it's win/loss ratio is

the number of winners/total trades or...

1273/1457=0.873713109=87%

can any other EA beat that I wonder?

To figure this out I isolated each position the EA took based on it's lot size. I could do that because this EA doubles the lot size for each trade in the series of trades it executes when it trades. I copied the strategy test results into excel and simply sorted for each trade of the same lot size and then counted how many won and how many lost for each of the five (when maxtrades=5) that it executes.

87% is pretty good I think. but I wonder if there are any better?

Anyone know of a better win/loss ratio?

that is awsome. But can it make real money. 87% is just cause to go live. Demo data feed from brokers are different. I am running CT on real account and demo too. Same exact settings but the EA is showing different decisions. One says soulution=buy other will say uncertian. Same broker, same time frame, same settings.

Dave

edit: with 87% winning on tests you'll know real fast if its working on real. If it takes more losses than what the test show the max loss is then disable it and cut losses. Worth a try.

 
xxDavidxSxx:
that is awsome. But can it make real money. 87% is just cause to go live. Demo data feed from brokers are different. I am running CT on real account and demo too. Same exact settings but the EA is showing different decisions. One says soulution=buy other will say uncertian. Same broker, same time frame, same settings.

Dave

edit: with 87% winning on tests you'll know real fast if its working on real. If it takes more losses than what the test show the max loss is then disable it and cut losses. Worth a try.

Those are good questions. I don't know why the differences occur that seem to plague testing environments.

Interestingly enough, the results got which are so good come from the time range of about Jan-Feb 2006 to the present. Prior to that going back to 2004.06.16 using alpari data the performance is very pathetic. If all I tested on was 2004-2005 I would have not given this EA a second glance, it's terrible. Then around first quarter 2006 some magic fairy waves a magic wand and this thing just catches hold of something...whamo..it goes into millions from nothing...It can start with less than $100 and still do it. I don't know what is different about the market or whatever that it doesn't perform until this year and then it goes stellar. I havn't as yet calculated the total win/loss ratio over the full three year peiord, so I just put forth the best of the best using only this years results.

I am wondering what I could do to make it still better. While I am considering going live with a mini account on it my wife made this statement, "If it really is giving this kind of growth then you have nothing to lose by being patient and allowing the demo forward test to teach you more about it before you risk real money."

 
furious_angel:
i have been using a more profit and loss system of measurements. i am sure that using both systems together would net some interesting results.

Tell me more.

 
 
can u tell us more about it

 

xyz if you're the ib.

 
 

so wouldnt you go with the Ea that does the same thing? so to speak, just like you sited earlier.. the ratio of winners to losers.. i hope this shed a different light on the subject.

 
Aaragorn:
I would go with ABC because they only lost 1/3 of what they made, whereas XYZ lost 4/5 of what they made.

I see that as ABC winning 66% and XYZ winning 20%.

Several assumptions here...All this is of course assuming that we are comparing apples to apples, meaning that they are both in the same industry group and working from the same basic inputs...or in the case of currencies... the same data over the same date range. So all other things being equal we are only looking at the output variances based on identical inputs, so that we are only measuring the difference in how they perform on the same conditions.

exacty right. You must consintrate on % not $ ammount.

its the draw down % and win/loss ratio that matters over time.

Also...05 had a different trading environment than 06. An ea that makes huge money in 50 won't in 06 and visa versa. But you can find a balance between the 2. If you can ballance an ea between 05-06 to give desirable %ages than you'll have an ea that will survive the test of time.

I belive I have done this with CT and am currently running it on real account.

Find the months with the worst results, make it give no more than a 38.2% draw down and then go back to the best months and make sure it still does good. An ea that will last years won't always give great profits each month. Just need it to limmit a draw down durring bad months.

Dave

123
New comment