Back/forward testing & Expectancy & Profit Factor

[syusuf66]

Hi all

So I had a question about expectancy/profit factor as well as back/forward testing if you could help. I have been paper trading a system for over 2 years (haven't been able to afford starting capital during this time) and recently started back testing it. My current trades taken and figures are as follows:

Back testing
90 trades taken
Profit factor is 1.34
Expected £ per trade (risking £100 per trade) - £8.76

Forward testing
76 trades taken
Profit factor is 1.24
Expected £ per trade (risking £100 per trade) - £8.16

I am planning to continue to forward test it for the year, but back test it until I have 200 trades in total then make a judgement on whether or not I should use it.

So my questions are:

- Are my current profit factors and expected £ per trade acceptable or should I be aiming for higher? I heard as close to 2 is best.

- Is 200 manually backtested trades enough to get a good enough idea of how the system will perform in live trading/forward testing?

It is worth noting that these figures are taking into account spread deductions too.

Any help would be hugely appreciated.

Thanks

 

[options-george]

Hi all
Thanks

Rather than taking a lot of calendar time to amass a large number of trades with your forward testing, would it work for you to 'forward-test' using a simulator such as ForexTester3? If so, you could complete your desired sample size in a matter of hours/days. I appreciate that your strategy may be such that it can only be forward tested in real time.

200 is an ok number, but generally the more the better. If you can complete a much larger number in a relatively short time, then definitely aim for more trades.

You should also ensure that you have tested strategy in different market regimes/environments.

Good luck.

 

[malaguti]

Hi all

So I had a question about expectancy/profit factor as well as back/forward testing if you could help. I have been paper trading a system for over 2 years (haven't been able to afford starting capital during this time) and recently started back testing it. My current trades taken and figures are as follows:

Back testing
90 trades taken
Profit factor is 1.34
Expected £ per trade (risking £100 per trade) - £8.76

Forward testing
76 trades taken
Profit factor is 1.24
Expected £ per trade (risking £100 per trade) - £8.16

I am planning to continue to forward test it for the year, but back test it until I have 200 trades in total then make a judgement on whether or not I should use it.

So my questions are:

- Are my current profit factors and expected £ per trade acceptable or should I be aiming for higher? I heard as close to 2 is best.

- Is 200 manually backtested trades enough to get a good enough idea of how the system will perform in live trading/forward testing?

It is worth noting that these figures are taking into account spread deductions too.

Any help would be hugely appreciated.

Thanks

Personally, a profit factor of even 2, let alone 1.3 is not something I'd be going live with.
If it helps, I have 2 strategies I use, one has a profit factor of 5 and the other, a factor of 9 based off 14 years worth of back testing

 

[syusuf66]

So your profit is 9 and 5 times your losses respectively for each system after 14 years of backtesting? Have you proved this in forward testing / live trading too? You’ll be a millionaire compounding that (dependant on your account size) for any sustained amount of time surely? Don’t mean to sound skeptical but that seems really high?

 

[malaguti]

So your profit is 9 and 5 times your losses respectively for each system after 14 years of backtesting? Have you proved this in forward testing / live trading too? You’ll be a millionaire compounding that (dependant on your account size) for any sustained amount of time surely? Don’t mean to sound skeptical but that seems really high?

forward testing also, live trading just for a few years
as for being a millionaire i'm following the plan and yes as you say that all depends on the starting capital. as for compounding there will come a point where i would not be trading as per the theoretical numbers. backtesting is of course all theoretical and is only a reference point.
only a fool would think that he/she could be buying hundreds of thousands shares in anything regardless of capital, so your compounding is also just hypothetical
but yes, it all looks lovely on an equity graph however im not trading an equity graph!
i'm just merely pointing out the possibilities

 

[syusuf66]

Fair enough. I’ve just seemed to read a lot of articles (not saying these hold anymore weight than what you are saying) that have said a profit factor closer to 2 is great, and anything above 3 is exceptional/un-heard of?

Have pasted one below.

http://tradingmarkets.com/recent/whats_your_profit_factor_heres_an_assignment_for_you-657965.html

Do you know why certain people would think a profit factor of close to 2 is acceptable and actually worth aiming for? I’m honestly not trying to challenge your opinion. If anything I’m trying to challenge myself and my own understanding of this by putting the questions out there..

 

[malaguti]

Fair enough. I’ve just seemed to read a lot of articles (not saying these hold anymore weight than what you are saying) that have said a profit factor closer to 2 is great, and anything above 3 is exceptional/un-heard of?

Have pasted one below.

http://tradingmarkets.com/recent/whats_your_profit_factor_heres_an_assignment_for_you-657965.html

Do you know why certain people would think a profit factor of close to 2 is acceptable and actually worth aiming for? I’m honestly not trying to challenge your opinion. If anything I’m trying to challenge myself and my own understanding of this by putting the questions out there..

I haven't read the article, and in honesty im probably not going to
I have no idea why certain people would think 2 is worth aiming for..possibly because they couldn't do any better and then decided to write an article claiming how great that number is? My beliefs are certainly not limited to the achievements of others
I hope the figure gives you now something worth aiming for?
if not, and you think 2 is great, why did you ask the original question? surely 1.34 is close enough to greatness for you?
whats your win loss ratio if you don't mind me asking?

 

[syusuf66]

I am undecided on whether or not 2 was great or not which is why I was asking in the first place. To see what other people aimed for and thought was acceptable.

My current win/loss ratio is on average around 35% win percentage, but I’m using a trend following system so not really concerned with this as it is about right for trend following systems.

 

[malaguti]

I am undecided on whether or not 2 was great or not which is why I was asking in the first place. To see what other people aimed for and thought was acceptable, and generally aimed for.

My current win/loss ratio is on average around 35% win percentage, but I’m using a trend following system so not really concerned with this as it isn’t about right for trend following systems.

That is quite poor, I’d certainly look to improve on that number.
Well look good luck to you. One thing though that could improve on your numbers is a more appropriate longer term timeframe

 

[syusuf66]

Poor according to who..?

 

[Bradtex]

In my experience trend following systems usually have much higher win/loss ratio. Usually it is the win ratio which is about 35% to 40%.

 

[malaguti]

Poor according to who..?

A good point, it is of course all relative. What’s great for one person..

 

[syusuf66]

A good point, it is of course all relative. What’s great for one person..

So is your profit factor on your live trades 9 as well? Or is this just from backtesting.

 

[Brumby]

Poor according to who..?

To your credit, you have nailed a very important point - what would be an objective benchmark. In the book "TRADING SYSTEMS, Secrets of the Masters by Joe Krutsinger, a total of 17 systems were featured that were shared by a number of traders including Michael Connors, Joe Dinapoli, Larry Willams et al. Summarised below are some of the key statistics from their systems which individually covered between 10 to 15 years of test data.

Unfortunately a key statistic missing is drawdown which in my opinion is the most important piece of data. You will noticed that not a single factor is the key driver but how the pieces come together that drives overall profitability.

 

[malaguti]

To your credit, you have nailed a very important point - what would be an objective benchmark. In the book "TRADING SYSTEMS, Secrets of the Masters by Joe Krutsinger, a total of 17 systems were featured that were shared by a number of traders including Michael Connors, Joe Dinapoli, Larry Willams et al. Summarised below are some of the key statistics from their systems which individually covered between 10 to 15 years of test data.

Unfortunately a key statistic missing is drawdown which in my opinion is the most important piece of data. You will noticed that not a single factor is the key driver but how the pieces come together that drives overall profitability.

Very well put Brumby, and thanks for sharing those stats from the book

 

[syusuf66]

To your credit, you have nailed a very important point - what would be an objective benchmark. In the book "TRADING SYSTEMS, Secrets of the Masters by Joe Krutsinger, a total of 17 systems were featured that were shared by a number of traders including Michael Connors, Joe Dinapoli, Larry Willams et al. Summarised below are some of the key statistics from their systems which individually covered between 10 to 15 years of test data.

Unfortunately a key statistic missing is drawdown which in my opinion is the most important piece of data. You will noticed that not a single factor is the key driver but how the pieces come together that drives overall profitability.

Thanks very much for this Brumby. That’s good to see.

One thing I’m confused about though how can a system like the one you highlighted (where the total profits are equal to 1.38 times the total losses) can be that much more profitable than the systems with higher profit factors? Obviously this shows (as you said) that profit factor should not be viewed on its own and as the sole stasticic, but how can this actually be the case in a practical sense? I’d imagine that if had a system that lost £1000 but made £1380 that would reflect this profit factor, but I can’t see how it could eventually reflect those type of returns...?

Or am I looking at this wrong?

 

[Brumby]

Thanks very much for this Brumby. That’s good to see.

One thing I’m confused about though how can a system like the one you highlighted (where the total profits are equal to 1.38 times the total losses) can be that much more profitable than the systems with higher profit factors? Obviously this shows (as you said) that profit factor should not be viewed on its own and as the sole stasticic, but how can this actually be the case in a practical sense? I’d imagine that if had a system that lost £1000 but made £1380 that would reflect this profit factor, but I can’t see how it could eventually reflect those type of returns...?

Or am I looking at this wrong?

There are two primary things to note.

1)It is about frequency of signals. If two systems are positive in expectancy and all things being equal, the one with a higher frequency of signals will be more profitable over a longer run. In other words, the one with a higher frequency in signals will be more efficient in utilisation of trading capital than the other one. Note the system that has a profit factor of 100 but only 2 signals in over a nine year period (not shown) basically would leave its capital idle for most of the time due to lack of signals.

2)Conceptually, a more robust trading system between systems is when one that has a higher average $ win/loss ratio than profit factor (all things being equal). This is connected to the issue of drawdown and number of consecutive losses and the one with a higher average $ win/loss ratio will suffer a lower drawdown for the same number of consecutive losses.

 

[syusuf66]

I see. That makes sense. Thanks for taking time to explain

 

[syusuf66]

There are two primary things to note.

1)It is about frequency of signals. If two systems are positive in expectancy and all things being equal, the one with a higher frequency of signals will be more profitable over a longer run. In other words, the one with a higher frequency in signals will be more efficient in utilisation of trading capital than the other one. Note the system that has a profit factor of 100 but only 2 signals in over a nine year period (not shown) basically would leave its capital idle for most of the time due to lack of signals.

2)Conceptually, a more robust trading system between systems is when one that has a higher average $ win/loss ratio than profit factor (all things being equal). This is connected to the issue of drawdown and number of consecutive losses and the one with a higher average $ win/loss ratio will suffer a lower drawdown for the same number of consecutive losses.

One other thing. By win/loss ratio, do you mean the percentages of winning/losing trades?

 

[Brumby]

One other thing. By win/loss ratio, do you mean the percentages of winning/losing trades?

No. Not percentage but $ ratio.