Importance of profit factor

To add comments, please log in or register
Mark Norris
34
Mark Norris  

Question to EA programmers out there with profitable, forward-tested EAs - what do you find to be a good profit factor (if you look at it at all)?


Obviously it has to be above 1, but I'm curious what you consider to be the sweet spot of profit factor? Do you shoot for something in the 1.5 range, something above 2 or even 3?


Any personal experience you've had with profit factor and forward testing would be great.


Thanks in advance.


Mark

Brian Dee - Random Trader
1759
Brian Dee - Random Trader  

MS

Drawdown amount is the first thing you look at - especially in relation to initial deposit

Then look at max consecutive losses in money

You are looking for <whats the worst this could do to me>

Assuming this is good quality live-account history from your own broker, then you look at PF, which should be at least 1.5 for a frequent trading EA (providing the StopLoss is not huge)

For a more typical EA, I would look for 2+ over a long period, personnaly I aim for 4, but this is unlikely to be sustained unless the EA switches off during periods of unsuitable market action

Keep in mind, the results in live trading will be less good - maybe by a large amount if your system is vulnerable to spread variation...

FWIW

-BB-

Mark Norris
34
Mark Norris  
BarrowBoy:

MS

Drawdown amount is the first thing you look at - especially in relation to initial deposit

Then look at max consecutive losses in money

You are looking for <whats the worst this could do to me>

Assuming this is good quality live-account history from your own broker, then you look at PF, which should be at least 1.5 for a frequent trading EA (providing the StopLoss is not huge)

For a more typical EA, I would look for 2+ over a long period, personnaly I aim for 4, but this is unlikely to be sustained unless the EA switches off during periods of unsuitable market action

Keep in mind, the results in live trading will be less good - maybe by a large amount if your system is vulnerable to spread variation...

FWIW

-BB-

Thanks BB. For some context what sort of systems do you normally trade (short timeframe scalpers, long-term, all of the above, etc.)?

Mark Norris
34
Mark Norris  
BarrowBoy:

MS

Drawdown amount is the first thing you look at - especially in relation to initial deposit

Then look at max consecutive losses in money

You are looking for <whats the worst this could do to me>

Assuming this is good quality live-account history from your own broker, then you look at PF, which should be at least 1.5 for a frequent trading EA (providing the StopLoss is not huge)

For a more typical EA, I would look for 2+ over a long period, personnaly I aim for 4, but this is unlikely to be sustained unless the EA switches off during periods of unsuitable market action

Keep in mind, the results in live trading will be less good - maybe by a large amount if your system is vulnerable to spread variation...

FWIW

-BB-

Also I'd be interested in your (or anyone's) critique of the following test results. Obviously the profit number is not going to pay for that private island but I'm more interested in reactions to drawdown, profit factor, etc. This EA is for EUR/USD, 5M. Test results are for the past 5 years.


Bars in test 369832
Ticks modelled 18337278
Modelling quality 90.00%
Mismatched charts errors 23
Initial deposit 100000.00
Total net profit 33235.39
Gross profit 106309.59
Gross loss -73074.20
Profit factor 1.45
Expected payoff 27.38
Absolute drawdown 196.69
Maximal drawdown 3934.00 (2.97%)
Relative drawdown 2.97% (3934.00)
Total trades 1214
Short positions (won %) 611 (87.07%)
Long positions (won %) 603 (86.57%)
Profit trades (% of total) 1054 (86.82%)
Loss trades (% of total) 160 (13.18%)
Largest
profit trade 120.65
loss trade -535.12
Average
profit trade 100.86
loss trade -456.71
Maximum
consecutive wins (profit in money) 40 (3970.02)
consecutive losses (loss in money) 2 (-1051.30)
Maximal
consecutive profit (count of wins) 3970.02 (40)
consecutive loss (count of losses) -1051.30 (2)
Average
consecutive wins 7
consecutive losses 1

Avi
52
Avi  

Looks amazing.

What is the time frame that this EA works on?

What is the test time (how many days/weeks/months)?

Avi
52
Avi  
Do you have a backtest graph?
Mark Norris
34
Mark Norris  

graph for the above data. As you can see the profits leveled off for the past 1.5 years (but the drawdown still remained OK. Any suggestions on improving the past 1.5 years?

Mark Norris
34
Mark Norris  
markelshark:

graph for the above data. As you can see the profits leveled off for the past 1.5 years (but the drawdown still remained OK. Any suggestions on improving the past 1.5 years?

also there is no LotSize/MM in this - this is v1 and is the same size lot for every trade. Obviously with MM in here the profit would be higher (but so would the drawdown).

Mark Norris
34
Mark Norris  
TuTBaluT:

Looks amazing.

What is the time frame that this EA works on?

What is the test time (how many days/weeks/months)?

5M, this is 5 year backtest.

samir dirgham
85
samir dirgham  
markelshark wrote >>

5M, this is 5 year backtest.

Hi markelshark

it is nice EA,

question: what is the reason that from trade#700 till the end of the testing period, the curve is straight (no gains) ?

can say 9 months

Mark Norris
34
Mark Norris  
samirgham:

Hi markelshark

it is nice EA,

question: what is the reason that from trade#700 till the end of the testing period, the curve is straight (no gains) ?

can say 9 months

Ok, it looks like I was able to filter out some of the bad trades from that time period. Here is what I did:


1. Ran dates from trade ~700 to end on VISUAL mode so I could eye exactly what was happening.

2. Added a technical indicator filter to filter out unfavorable market conditions

3. Re-optimized for best inputs on the technical filter.

4. Chose results with lowest drawdown and highest profit (basically same profit as above but signifcantly less draw down).

5. re-ran tick test


It comes out looking pretty good. I ran 10-year data too and the curve stays true.

12
To add comments, please log in or register