Great EA in backtest! - page 70
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
Does anyone know anything about the Exclusive Forum? Do they have access to "superior" EA's or other information?
Yes there is some good stuff in the Elite section, it's worth the price, but CT is more active right here in this thread at the moment.
Some month ago I have tested 3 or 4 experts from a russian site. These experts had a self learning component. The first backtest was bad, the second better and the third one was good. These experts stored a .ini file or something like this in MT. After deleting this file the expert was stupid again. Unfortunately I don't know the side of the programmers but if this would be good for devellopement of CT I can search it again.
CT uses neural learning to make descisions, yes its needs to run for a period of time and not be interupted, its also best not to run any other EA's in the same MT4 session as CT.
The author of CT has begun putting together Neural netwrok database that the Pro version will connect to, this allows everyone running the EA to contribute to the simulation learning patterns and to read from them instantly as well.
Also note that since CT uses Neural simulations to make descisions, the indicators that you enable only add "influence" to CT, CT takes your buy/sell input and will perform a simulation around that influence. So if you say RSI to Sell when price drops down thru 70 CT will begin to simuate a trade at that sell point, but it doenst care about RSI at all, just that you gave it input that you'd like to sell at that point.
more evidence of this is if you take 2 MT4 setups that are identically setup with CT, and start them up only minutes apart they will trade differently since thier simulation/learning time is also different...this is at least best I can figure why 2 setups or more will all trade differently using the same price feeds and settings... hence possible why demo and live dont always trade the same
Something I see keeps coming up is that ppl expect to see a profit right away, you need to be patient, this EA needs to run a couple weeks or months before you says its good or bad... My testing says it has great potential, but I'll only say its a success after more than 3 months running.
last week was the first profitable week running out of about 3 1/2 weeks, but its also the first week I left it alone and let it run with NO interuptions.
Back testing hasnt done much justice for this EA except to confirm that it can actually make trades, every ones tests seem to be dramatically different.
I've forward tested most all the versions live and only 185f has worked successfully. 188 sucked the $ outta my account very quick, 185f got some of it back for me and is slowly growing upward
"I am by no means an expert on this EA, I'm just sharing my observation from my live use of this EA, I am STILL learning from this EA.. I could be wrong about some this"
Here you go. . . it's also on post #333. I also added 185f2 which I haven't had a chance to test yet.
thanks DIVA!!!!
What TIME FRAME have you had the best results???
cheers
thanks DIVA!!!!
What TIME FRAME have you had the best results???
cheersI've been using the 30M.
I've been using the 30M.
ok,
and I guess on the recomended pairs of EURUSD and USDJPY
thanks
CT uses neural learning to make descisions, yes its needs to run for a period of time and not be interupted, its also best not to run any other EA's in the same MT4 session as CT.
The author of CT has begun putting together Neural netwrok database that the Pro version will connect to, this allows everyone running the EA to contribute to the simulation learning patterns and to read from them instantly as well.
Also note that since CT uses Neural simulations to make descisions, the indicators that you enable only add "influence" to CT, CT takes your buy/sell input and will perform a simulation around that influence. So if you say RSI to Sell when price drops down thru 70 CT will begin to simuate a trade at that sell point, but it doenst care about RSI at all, just that you gave it input that you'd like to sell at that point.
more evidence of this is if you take 2 MT4 setups that are identically setup with CT, and start them up only minutes apart they will trade differently since thier simulation/learning time is also different...this is at least best I can figure why 2 setups or more will all trade differently using the same price feeds and settings... hence possible why demo and live dont always trade the same
Something I see keeps coming up is that ppl expect to see a profit right away, you need to be patient, this EA needs to run a couple weeks or months before you says its good or bad... My testing says it has great potential, but I'll only say its a success after more than 3 months running.
last week was the first profitable week running out of about 3 1/2 weeks, but its also the first week I left it alone and let it run with NO interuptions.
Back testing hasnt done much justice for this EA except to confirm that it can actually make trades, every ones tests seem to be dramatically different.
I've forward tested most all the versions live and only 185f has worked successfully. 188 sucked the $ outta my account very quick, 185f got some of it back for me and is slowly growing upward
"I am by no means an expert on this EA, I'm just sharing my observation from my live use of this EA, I am STILL learning from this EA.. I could be wrong about some this"Thank you for your insights, this is really helpfull for me.
Just give you additional information. The CT doesn't work well with built 196 and 197. It works best with built 195. So guys, do not upgrade it. Just use built 195.
hi fikko,
still the same problem. mt build 195. results attached.
CT uses neural learning to make descisions, yes its needs to run for a period of time and not be interupted, its also best not to run any other EA's in the same MT4 session as CT.
hi DudeWorks,
are You talking about OS version of CT?
i've done a small experiment using OS version. after backtest it saved all variables values to a file. then i started another test (with variables initialized with those previously saved values) - same backtest results. so in my opinion OS version does not learn.
EDIT: another thing is that OS version deos not save any stats in a file. so if one stops CT for a weekend it'll loose all that has learned if there is learinig process. i can't imagine running CT for a month or two without even closing mt for a moment.
hi DudeWorks,
are You talking about OS version of CT?
i've done a small experiment using OS version. after backtest it saved all variables values to a file. then i started another test (with variables initialized with those previously saved values) - same backtest results. so in my opinion OS version does not learn.
EDIT: another thing is that OS version deos not save any stats in a file. so if one stops CT for a weekend it'll loose all that has learned. i can't imagine running CT for a month or two without even closing mt for a moment.Like I said, i could be wrong, just my impression.. but you said you saved the variables that it had produced .. basically saving its state... then start another session with those variables..isnt that passing its previous state to itself? ... a true Neural network requires a database to store what it learns so it can keep building on that, CT cant really do that, it only keeps the last pattern assumptions in memory.. which you stored in your file...so that woud be like pausing it for a moment... but I guess that doesnt really explain your back test results either...
You can see the CyberiaLogic do simulations and places scores on the simulations, so its definately doing some sort of learning..
appears to do simulations for a period of time
examples:
2006.10.09 14:13:28 Cyberia 185f EURUSDm,M15: Period of the simulation:3 minutes with the probability:0.5
2006.10.09 14:22:23 Cyberia 185f EURUSDm,M15: Period of the simulation:7 minutes with the probability:0.625
Have you been able to get a back test to match a forward test?
Have you been able to get a back test to match a forward test?
i didn't try, because it's only me who gets so optimistic results.
i'll optimize sl for 4 months period and will run ct on real account till friday. we'll see.