4 System Rules - Which Would You Choose?

[daedalus 10]

I think this is the appropriate forum for this kind of question.

 

I am developing a new method and have some backtesting results done which were then filtered 4 different ways.

 

My question to you is which of these outcomes would you consider the best?

 

 

My stumble point has been the trade efficiency vs. overall $ profit per contract. I don't know which one is best...

 

I like high efficiency but passing up on essentially double the profit (net of commissions) seems like a somewhat stupid idea...

 

Any thoughts?

[gooni 10]

Profit is only a function of contracts traded. I'd like for highest efficiency or highest win rate or lowest drawdown (and at best, all three) - because the goal is to increase leverage and these are the factors that make that step the easiest and least psychologically challenging.

 

Although the number of trades for all systems looks far too low to be drawing a conclusion from, especially something like efficiency.

[UrmaBlume 10]

I think this is the appropriate forum for this kind of question.I am developing a new method and have some backtesting results done which were then filtered 4 different ways.My question to you is which of these outcomes would you consider the best?My stumble point has been the trade efficiency vs. overall $ profit per contract. I don't know which one is best...I like high efficiency but passing up on essentially double the profit (net of commissions) seems like a somewhat stupid idea...Any thoughts?

 

IMHO 30 or less round trips in not enough of a sample space to produce dependable results.

 

The reliability of such tests increases, not with the passage of time, but, rather, with an increase in the number of transactions. In one of your examples you show less than 10 transactions and that's not enough to count on for anything, IMHO.

[symmetrics 10]

agreed min 6mnths testing , i prefer 2 years or 5 and make sure you have a bull move and bear move , period , otherwise when your system falls apart on a given change your system goes out the window , then so do you so to speak :crap:

[daedalus 10]

understood on the validity of anything statistically significant on these amount of trades...

 

That being said... say each had been produced over 1000 trades....

 

which would you choose then - higher efficiency and less profit or higher profitability with more trades?

[daedalus 10]

I don't know if this would make a difference but here are the two of the equity curves...

 

The first attachment is for Outcome 1, the second attachment for Outcome 2.

 

As you can see both have minimal drawdowns but Outcome 2 has a decidedly more upwards and direct equity curve.

[symmetrics 10]

ok

 

brokerage and slippage needs to be defined , your focus on profits is not the only thing to take into the equation

 

so here goes - ideally as said before data needs to be sited thru periods of up and down markets. that said the your question on the options of trades is this.

 

traders need to focus on risk to reward and drawdowns , your profit is really irrellavant to trades as soem factors such as leverage, brokergae rates should be extremely low, so your greatest factor wil be slippage ( positive or negative ) your risk on the trade ( stops or no stops ) and profit vs. drawdowns.

 

volitilty is a important factor too as knowing when to now allaocate alot to mininal amounts

 

your question is too broad as in one way as your system needs to accomadate managment of funds and when to pile onthe contracts and when not too

 

i guess to have a fair assesment id need to measure some of these factors, management of tades is jsut as important as profit , at the end of the day oyu can the most profitbale system yet loose the lot on alaoction or bad management of it , look at larry wiliams 87 trades he really made 4mill plus but gave it up in the last 3 mnths due to bad management , yes he still did 10,000% but wiped 300 % in 3 mnths - so the bottom line trades dont matter much, serious traders pay next to nothing for brokerage so id go with profit , low draw downs and many trades. ( in this case basedon whay limited detials ive seen)

 

thanks :missy:

[BlowFish 10]

I would want to know the typical Risk Reward too. I would tend to go with the highest % winners though if you where risking 4 to make 1 that might give me second thoughts. My guess is the last system is likely to have that sort of RR. High percentage systems allow you trade larger size with similar risk of ruin, however I simply find them easier to execute.Of course people want to risk 1 to make 10 and get 80% winners, aint gonna happen.

 

It would be good to know the typical holding period and how long in the market each is. Something that trades 4 times a year and is always in might not suit many.

 

I am not a systems trader but half the thing is to know when they are trading atypically the more information you have about expected performance the more you can tell when a system is deviating.

[kappert 10]

Have been asking myself similar questions. Without any theory to back this up, just common sense:

 

I use the results of my optimizations with the best total net profit. When I see that they at same time produced the best efficiency, I know that I am on to something. When optimization results with best total net profit and best efficiency differ widely, I become hesitant.

 

More trades means more work (unless it is an automated system). Could you actually perform all these trades in real life with the correct timing?

Of course you should factor in transaction costs. 49 trades may cost significantly more than 15 trades. This can turn the whole result around.

 

And I agree: Longer data series are required for meaningful back-testing.