Profit Factor Phase Space | Forex Factory

Options

Permalinks
Print Thread

Similar Threads

Poll for your Profit Factor 46 replies

New Expert Advisor PROFIT FACTOR 19.39 47 replies

Profit Factor OR Expected Payoff...Which is more important?? 5 replies

Profit Factor Target 27 replies

Profit factor : 11.64 2 replies

Attachments: Profit Factor Phase Space
Exit Attachments

Profit Factor Phase Space

  • #1
  • First Post: Edited 4:51am May 2, 2010 3:07am | Edited 4:51am
Before I go into the topic, I would like to highlight that this thread is not going to discuss any trading systems but the scientific method to exploit the optimal profit potential of a quantitative trading strategy.

Like many of you here, I suspected that there could be an optimal profit factor for a systematic trading system. I have done a lot research and I came out this one, the profit factor phase space.

Here are the sample profit factor phase space for my quantitative model. Both charts are the same but at different angle of view. Btw spread is not taking into consideration in these charts so that i can minimize the external interference that distort the true statistical behaviour of my model.

I just wonder any of you did a similar study on your trading strategy? Let's talk.
Attached Image(s) (click to enlarge)
Click to Enlarge Name: PF_rotation20deg.PNG Size: 56 KB
Click to Enlarge Name: PF_rotation200deg.PNG Size: 56 KB
  • #2
  • May 2, 2010 4:46am May 2, 2010 4:46am
i haven't come accross this kind of study in forex trading but i'm all ears
Success consists of going from failure to failure without loss of enthusiam
  • #3
  • Edited 6:07am May 6, 2010 5:49am | Edited 6:07am
Hi ForexQuant,

I may have tried to do something similar, but I am not sure. I've tried to understand your chart, but I don't know what the "scale parameter" on your chart refers to (it isn't RRR, when I check the shape), and so failed to understand it.

cheers,


Quoting ForexQuant
Before I go into the topic, I would like to highlight that this thread is not going to discuss any trading systems but the scientific method to exploit the optimal profit potential of a quantitative trading strategy.

Like many of you here, I suspected that there could be an optimal profit factor for a systematic trading system. I have done a lot research and I came out this one, the profit factor phase space.

Here are the sample profit factor phase space for my quantitative model. Both charts are the same but at different angle of view. Btw...
  • #4
  • May 6, 2010 9:36am May 6, 2010 9:36am
Very interesting.

Could you tell us on how many trades your graph is based. I am finding that profit factor decreases the more trades there are. This does not mean it isn't profitable. Have you come to that conclusion?
  • #5
  • May 6, 2010 10:50am May 6, 2010 10:50am
ForexQuant,

Looking forward for more background...
  • #6
  • May 6, 2010 11:14am May 6, 2010 11:14am
No Brainer Trades OP (a thread on forexfactory) may have done a similar analysis for his trading method as ForexQuant, although he did not make a 3-dimensional surface chart.

http://www.nobrainertrades.com/2010/...placement.html
  • #7
  • May 6, 2010 2:25pm May 6, 2010 2:25pm
Quoting kk007
Hi ForexQuant,

I may have tried to do something similar, but I am not sure. I've tried to understand your chart, but I don't know what the "scale parameter" on your chart refers to (it isn't RRR, when I check the shape), and so failed to understand it.

cheers,
Usually probability distribution has 2 parameter. One is scale parameter and another one is location parameter.
  • #8
  • May 6, 2010 2:33pm May 6, 2010 2:33pm
Quoting spekkiefx
Very interesting.

Could you tell us on how many trades your graph is based. I am finding that profit factor decreases the more trades there are. This does not mean it isn't profitable. Have you come to that conclusion?
I plotted this graph based on probability distribution of my model. It is not based on the trade results. However the model was built to fit the trade results.

IMO the profit factor does not "decrease", it is just approaching its true profit factor.
  • #9
  • May 6, 2010 4:12pm May 6, 2010 4:12pm
Quoting ForexQuant

IMO the profit factor does not "decrease", it is just approaching its true profit factor.
Nicely formulated (-:
  • #10
  • May 7, 2010 12:23am May 7, 2010 12:23am
We all know that the external factors like spread & slippage, etc play an important roles in our profitability. But how they screw us? Let's see.

Here are the charts for ideal case and non-ideal case for my model. For non-ideal case, I have included only 1pips for spread & slippage. Now you see the shape of the surface is completely different on the left side of the chart!
Attached Image(s) (click to enlarge)
Click to Enlarge Name: Ideal PF.PNG Size: 53 KB
Click to Enlarge Name: Nonideal PF.PNG Size: 53 KB
  • #11
  • Edited 1:27pm May 7, 2010 1:24pm | Edited 1:27pm
ForexQuant,
Can you please give a definition for the profit factor ?
the profit factor is made from the %win and the risk/r ,but the %win is also use as a parameter ,so i am a bit confused.(from : http://www.forexfactory.com/showthread.php?t=222748)
(I don't want to troll or anything it's look intressting,so i want to be sure of what i understand )
“Aim for the highest.” Andrew Carnegie
  • #12
  • May 7, 2010 1:47pm May 7, 2010 1:47pm
Quoting Tenetch
ForexQuant,
Can you please give a definition for the profit factor ?
the profit factor is made from the %win and the risk/r ,but the %win is also use as a parameter ,so i am a bit confused.(from : http://www.forexfactory.com/showthread.php?t=222748)
(I don't want to troll or anything it's look intressting,so i want to be sure of what i understand )
profit factor = %win / % loss * Reward/Risk;

No i didnt use %win as a parameter. You can convert %win to Reward/Risk by using the standard profit factor formula.

I prefer to use %win because %win always fall within 0 - 100% therefore it is easier to make a comparison for multiple pairs.
  • #13
  • May 7, 2010 1:56pm May 7, 2010 1:56pm
okay thanks
“Aim for the highest.” Andrew Carnegie
  • #14
  • May 7, 2010 10:40pm May 7, 2010 10:40pm
My model has a significant positive skew distribution therefore a fat tail in nature. Before I created the phase space diagram for analysis, I thought that the profit factor should be approaching infinity if I use extreme reward/risk (Low win%). However this diagram has prove me wrong!

My model has a relatively stable scale parameter, which is about 1.8 +- 1 (I dont know why) for multiple pairs, multiple setting. Now we look at the attached phase space diagram again. Look at the red line, what you see? Yes it has a downward slope, which means the profit factor does not increase as i original thought! Instead It has an optimal profit factor around 15-20% zone!

Interesting right?
Attached Image (click to enlarge)
Click to Enlarge Name: optimal pf.PNG Size: 53 KB
  • #15
  • May 8, 2010 5:35am May 8, 2010 5:35am
I have some figures but haven't had the opportunity to put them in the charts.
I do have a feeling the results may be similar. I was quite surprised by the massive effect of 1 pip slippage chart on the high percentage winning systems. I am not sure if that is accurate. What type of data are you using? Or maybe it is specific to the system you are testing.
  • #16
  • Last Post: May 9, 2010 12:15am May 9, 2010 12:15am
Quoting spekkiefx
I have some figures but haven't had the opportunity to put them in the charts.
I do have a feeling the results may be similar. I was quite surprised by the massive effect of 1 pip slippage chart on the high percentage winning systems. I am not sure if that is accurate. What type of data are you using? Or maybe it is specific to the system you are testing.
I believe that the massive effect on the left is the behaviour of my setup because the scale parameter is about 1.8. If i can find a way or a setup which can reduce the scale parameter to 1.5 or something lower, the 1 pip effect on the left will be insignificant as per attached chart.

However, when the pip size increases gradually, the surface begins to deform and eventually back to the shape that i posted at post#10. The left side of the surface has a big deformation while the right side of the surface will remain similar to the ideal case.

I am not sure about other type of systems but I can conclude that the high probability setup of my model is vulnerable to external shock in regardless of scale parameter. The unforeseen event like widening spread, huge slippage, negative blackswan event or change in market dynamic will cause a great impact and maybe completely unpredictable if i do high probability trade.
Attached Image (click to enlarge)
Click to Enlarge Name: Suface with 1pip Spread.PNG Size: 58 KB
0 traders viewing now