Increase Variables allowed in Optimization?

 

I have recently written some fairly complex EA's and when I attempt to use the optimizer I all too often get the Error message in my journal: "Too many passes for optimization, please try to decrease optimized parameters or to increase parameters' values step."


I was just curious is there is a way arround this.

Any ideas? Thanks for your time and patience you guys are a great community!!

 

CCG

1) Not good to optimize too many parameters at the same time – the probability of fitting for history will be too high.

2) It takes ages to run with the many, many permutations

3) To be direct, it sounds like you dont know what its likely ranges are?


If its a 'programmers EA' then the narrowing down of ranges should be done as the EA grows in stages. (Says BB who has done some 100 hour runs!!)

If its a coded-up trading system EA, then common sense main values should work, eg if its opening/closing on an M1 chart, you probably dont need a TP=1000!

I usually develop initial approximate TP/SL/TS settings, then permutate the inner trigger values to get a narrow range of those.

Then put narrow ranges of *.* variables in a final optimisation run on single-broker (not MT) data

Good luck with it tho!

My 2c worth

-BB-

 

Well My intent is to find the optimal settings on each light the indicator is written to run and work on open prices.

The dilemia I run into is that the 4h light could run anywhere from 100-1000 Take Profit, the Daily period just the same.

However that is hardly one of the many variables.

This system is extremly flexible and can work from 15 - 1000 TP its not bound to any particular settings due to the fact it is based off of compression of SMA's, SMA direction, And another intricate indicator that limits all the trades to the "Perfect setup"

This is fully functional and it does work properly without any serious optimization I'm just looking to find that sweet spot.

I suppose I will do some research on how to write a program that can optimize it without restricting it to so few variables. 1Trillion variables is just too few, Haha.


Does anyone know how well the "auto optimizer" written by that russian fellow works?


Thanks for the help again.

 

Thanks for the advice I will procede in this manner.

BarrowBoy:

CCG

1) Not good to optimize too many parameters at the same time – the probability of fitting for history will be too high.

2) It takes ages to run with the many, many permutations

3) To be direct, it sounds like you dont know what its likely ranges are?


If its a 'programmers EA' then the narrowing down of ranges should be done as the EA grows in stages. (Says BB who has done some 100 hour runs!!)

If its a coded-up trading system EA, then common sense main values should work, eg if its opening/closing on an M1 chart, you probably dont need a TP=1000!

I usually develop initial approximate TP/SL/TS settings, then permutate the inner trigger values to get a narrow range of those.

Then put narrow ranges of *.* variables in a final optimisation run on single-broker (not MT) data

Good luck with it tho!

My 2c worth

-BB-

Reason: