It's possible someone create a profitable EA that operates without backtesting and without any input parameter?

New comment
 
  • 20% (17)
  • 21% (18)
  • 6% (5)
  • 39% (33)
  • 13% (11)
Total voters: 84
 
Imagine a profitable robot that operates without backtesting and without input parameters. Dream, fiction or reality? 

Anyway, leave a comment if you wish, stating what was your answer and her logic.
 
I think anything is always possible. It can happen by well planning strategy or may be by "accident" :)
 

Probably Yes?

The question should be how many of you have made a profitable EA?

Followed by how many made this profitable EA without Back-Testing?

That should give a clear answer to whats Probable.

Anything is possible. Like time-travel for example :) 

 

Without input parameters means self-adaptative.

Without backtesting means self-backtesting.
A bot able to simulate itself different strategies and evaluate the performance.

Yes, is possible.
I guess they already exist.

Obviously every system has its own limits.

 
figurelli:
Imagine a profitable robot that operates without backtesting and without input parameters. Dream, fiction or reality? 

Anyway, leave a comment if you wish, stating what was your answer and her logic.
Backtesting is not just about performance evaluation it is also about software testing.
 

probably yes.

the EA that I used in ATC 2012 is using a intra day candle pattern

the input parameter range to optimize is very limited

in fact when I finish coding, i tested straight in the strategy tester and its profit (without optimization).

 
RaptorUK:
Backtesting is not just about performance evaluation it is also about software testing.

certainly. few days back when I tested my EA, I'm running optimization and just the next parameter blows the account.

when I investigate, it was a coding problem. 

 
I have :) one EA applied to one chart trading all symbols in market watch widows.  It earned over 600 USD in less than 10 hours trading only 0.01 lot sizes on 33 symbols.  I'm still developing it but it works great in demo already :) no optimization whatsoever.  And it wasn't "luck" it was just well thought out!
 
johnnypasado:
I have :) one EA applied to one chart trading all symbols in market watch widows.  It earned over 600 USD in less than 10 hours trading only 0.01 lot sizes on 33 symbols.  I'm still developing it but it works great in demo already :) no optimization whatsoever.  And it wasn't "luck" it was just well thought out!
Thanks for your contribution, but this is not a short time to a conclusion? Anyway, it would be nice to have a white box and open source example to we study here. 
To be honest, I see some ways to do this with several EAs, but I would like to get the same way with just one 'magic' EA.
 

that is the only way you can build an EA that works consistently.  it has to be non-parametric, as the nature of price action/movement is constantly changing. it should depend only on one pass statistical analysis of shorter history to get 'data relation', as it is not possible to remove all parameters, i.e. some relationships need to be defined.

all other systems have expiration date, usually a one that will eat all profit you made until you realise that rules have changed.

and yes, i am also going in that direction and i have some stuff that is working in that sense. not completely clean, but getting closer and closer. 

1234
New comment