Test ur strat

cr6196 · Jun 24, 2009

I am about to start backtesting my new idea/strategy/theory whatever...and thought i would try and use this resource to get some new ideas as what to look for etc.

Not reli new to backtesting but haven't found anything with real positive expectancy yet...

so when testing your entry criteria, what are the most useful things to look at...i usually look at how many points then entry goes against me and how far it goes in the "right direction"...any other useful things to look at here?

i would also add i am not gifted with any particular math skills.

Technically Fundamental · Jun 24, 2009

I have no idea how to write a strategy (automated) or backtest. I just watch the charts. feel like a bit of a dunce tbh. I'll be following this.

jiggly · Jun 24, 2009

cr6196 said:
so when testing your entry criteria, what are the most useful things to look at...i usually look at how many points then entry goes against me and how far it goes in the "right direction"...any other useful things to look at here?

i would also add i am not gifted with any particular math skills.

how many points it goes against you before coming good should not really be your concern, unless it is taking you out just before shooting off in the opposite direction. you should then be looking to hone your timing.

for an entry you should primarily be concerned with, how much you risk, based off of the criteria which gets you into the trade and the probability that the market will move in your favour.

exits are more important than entries

the journey is not as important as the destination..........

cr6196 · Jun 24, 2009

Ok ill add a bit more...

strike rate seems to be the most obvious omission i've made when testing the validity of the entry. What other tests should you run on the idea though? no.of consecutive losses? etc.

but how many trades should you go back and test through? i was thinking of testing my strat on 1h? so 50? 75? 100?

cr6196 · Jun 24, 2009

jiggly said:
how many points it goes against you before coming good should not really be your concern, for an entry you should primarily be concerned with, how much you risk, based off of the criteria which gets you into the trade and the probability that the market will move in your favour.

the journey is not as important as the destination..........

but how can u know how much to risk? without doing this it seems like there isn't any other way to statistically confirm ur stoploss and then set ur risk based on that?

how would you test probability? no.of times it fails/works?

jiggly · Jun 24, 2009

cr6196 said:
Ok ill add a bit more...

strike rate seems to be the most obvious omission i've made when testing the validity of the entry. What other tests should you run on the idea though? no.of consecutive losses? etc.

but how many trades should you go back and test through? i was thinking of testing my strat on 1h? so 50? 75? 100?

this completely depends on you,

I would say you should test your strategy on as much sample data as you can get your hands on, or a sample of 100 trades will give you a pretty good idea as to how it will perform.

from your results, you will be able to work out,

win vs loss ratio,
average, max and min risk vs reward,
equity drawdown,
average return......
etc....

jiggly · Jun 24, 2009

cr6196 said:
but how can u know how much to risk? without doing this it seems like there isn't any other way to statistically confirm ur stoploss and then set ur risk based on that?

how would you test probability? no.of times it fails/works?

you should know how much you are comfortable with losing before placing the trade. e.g 2% of equity.

from there, you will calculate, at what point the reason to place the trade has become invalidated, i.e, if you see a head and shoulders pattern and you are entering on the break of the neckline, your stop should logically be placed, either just above the shoulder or above the head.

the event that got you to initiate the trade was the head and shoulders pattern, and your trade will be invalidated if the pattern does not play out as anticipated.

you know the distances between your entry and when the pattern will be invalidated, so carry as much stock as your risk allows you to.

jiggly · Jun 24, 2009

cr6196 said:
how would you test probability? no.of times it fails/works?

precisely

cr6196 · Jun 24, 2009

jiggly said:
you should know how much you are comfortable with losing before placing the trade. e.g 2% of equity.

from there, you will calculate, at what point the reason to place the trade has become invalidated, i.e, if you see a head and shoulders pattern and you are entering on the break of the neckline, your stop should logically be placed, either just above the shoulder or above the head.

the event that got you to initiate the trade was the head and shoulders pattern, and your trade will be invalidated if the pattern does not play out as anticipated.

you know the distances between your entry and when the pattern will be invalidated, so carry as much stock as your risk allows you to.

interesting, made me think about if i am trading a pattern and where it wud be invalid etc.

i think you can only say how much risk u shud take once u r confident in ur win/loss ratio, consec. win/losses etc. but that might just be me...

can any1 think of anything missed here?

jiggly · Jun 24, 2009

cr6196 said:
interesting, made me think about if i am trading a pattern and where it wud be invalid etc.

i think you can only say how much risk u shud take once u r confident in ur win/loss ratio, consec. win/losses etc. but that might just be me...

can any1 think of anything missed here?

how much to risk, is completely down to the individual and the tolerance for 'risk'...

the £ amount you will risk per point, will vary, dependent upon your timeframe and the setup, but the total amount at risk should be known before hand, for each and every trade your take.

Test ur strat

cr6196

Well-known member

Technically Fundamental

Senior member

jiggly

Well-known member

cr6196

Well-known member

cr6196

Well-known member

jiggly

Well-known member

jiggly

Well-known member

jiggly

Well-known member

cr6196

Well-known member

jiggly

Well-known member

Similar threads