Intelligent algorithms based on behavioral theory and pattern recognition ... - page 3

To add comments, please log in or register
Ubzen
5299
Ubzen  
abstract_mind:

In this regard, and not trying to steal your thread because this might also be relevant for your trading strategy, I would like to ask opinion to all guys about a realistic value for the "Profit Factor", e.g. for long term resilient EAs (not tuned/optimized with short term settings).

My_Straight_Answer= Profit_Factor=1.5

My_Opinions: Profit_Factor >=2, spanning years = Holy Grail.

My_Note_About_PF: "The more profit you tell it to take, the higher the profit factor will be"

Example: Grid strategy, lets say 3 levels.

------Buy1

------------Buy2

-------------------Buy3

if you close the orders out when price when back-up to level 2 (slightly above break-even) then you'll have lower PF then if you close orders out when they come back to level 1 or higher. **Hint: this is how you increase profit factors to 5+. Warning: this does not make grid trading safer, nor does it mean a system with PF 7.1 is a whole lot better than a system with 1.1. <--- Same can be said for someone who have Big take profit and Small stop_loss.

All in my Opinions.

dabbler
1091
dabbler  
ubzen:

My_Straight_Answer= Profit_Factor=1.5

My_Opinions: Profit_Factor >=2, spanning years = Holy Grail.

My_Note_About_PF: "The more profit you tell it to take, the higher the profit factor will be"

.. nor does it mean a system with PF 7.1 is a whole lot better than a system with 1.1. <--- Same can be said for someone who have Big take profit and Small stop_loss.

All in my Opinions.

Well this is worthy of discussion.

As far as I am concerned changing the take profit up doesn't ordinarily affect the profit factor in the long term. A system with a profit factor of 7 over an adequate number of trades (>40 say) has plenty of lee-way on errors whereas a system with a profit factor of 1.1 is untradeable unless that is based on thousands of trades back or forward tested on real tickdata.

Metrics on grid trading are more difficult because when trading without a stop it is the maximum open drawdown that is the critical factor.

Joao Rosas
391
Joao Rosas  
ubzen:

My_Straight_Answer= Profit_Factor=1.5

My_Opinions: Profit_Factor >=2, spanning years = Holy Grail.

My_Note_About_PF: "The more profit you tell it to take, the higher the profit factor will be"

Example: Grid strategy, lets say 3 levels.

------Buy1

------------Buy2

-------------------Buy3

if you close the orders out when price when back-up to level 2 (slightly above break-even) then you'll have lower PF then if you close orders out when they come back to level 1 or higher. **Hint: this is how you increase profit factors to 5+. Warning: this does not make grid trading safer, nor does it mean a system with PF 7.1 is a whole lot better than a system with 1.1. <--- Same can be said for someone who have Big take profit and Small stop_loss.

All in my Opinions.


Thank you Ubzen. :-)
Ubzen
5299
Ubzen  
dabbler:

Well this is worthy of discussion.

As far as I am concerned changing the take profit up doesn't ordinarily affect the profit factor in the long term. A system with a profit factor of 7 over an adequate number of trades (>40 say) has plenty of lee-way on errors whereas a system with a profit factor of 1.1 is untradeable unless that is based on thousands of trades back or forward tested on real tickdata.

Metrics on grid trading are more difficult because when trading without a stop it is the maximum open drawdown that is the critical factor.

I agree with your assessment 99%. Except for the 41 trades being meaning-full for high profit-factors (Just for Devil's Advocate 8P). Of-Course the one with PF=7 is superior to PF=1.1 when all else consider equal. But they're never all else equal / nor even close. If someone's comparing 1.1 to 7 then the #-of-trades on 1.1 is probably higher or there's no contest.

The conventional wisdom I read about all over the web when I started Trading went something like make sure you have a 2:1 TakeProfit vs StopLoss that way, you only need X wins to recover from Y losses. It all sounds good, and someone would draw the conclusion that trading with (1000_Tp & 10_Sl) is better than (Tp_10 & Sl_1000) but neither of the two is better. Well until you start to consider the Win%.

Generally, the story with 1.1 is that he goes months winning and then one day Blow-the-entire-account. Generally, the story with 7.0 is that he goes down slowly. That 41 trades was just a snap shot because of his Low# of trades for the tested Year. Give him 10 years back-test and you'll start to see flat (at best) ----- to downward equity curves for 2-3 years in a roll which never fully recovers. No full-time trader can trade this.

So either way, you end up wasting time with PF=7 or Money with PF=1.1. This is just my Opinion about Justifying Low#of trades <1000 because the PF is High. Any Statistic by itself without considering others like Drawdown, Win%, #-of-Trades etc is meaningless.

123
To add comments, please log in or register