Mechanisation of optimal parameter selection. Finding a common denominator. - page 4
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
and if you don't see the profit factor in the tester at all - there is no loss... it has long been agreed not to divide by zero ... what should you "take" next?
it means few trades and no statistical validity (faith in the results)
Yeah... so the number of trades comes before the profit factor... but we can assume the system is capable of closing orders piecemeal and some positions will be trawled and closed with stops... and the number of trades is not reflected on the statistical validity... or rather, it does, but it does not...
Yeah... so the number of trades is higher than the profit factor... but we can assume the system is capable of closing orders piecemeal and some positions will be trailed and closed with stops... and the number of trades is not reflected on the statistical validity... or rather, it does, but it doesn't rule...
Yes, the number of trades is an important parameter. For example it may be a good enough number of deals, but it does not increase the reliability of your stats. It starts with dependent trades (for example one entry can be divided into 10 equally strong ones and saddleloch will be 10 times bigger) and finishes with such nuances as ratio of average profit trade to average loss trade (for systems where profit trade is larger than loss trade and vice versa we need much more trades than when they are approximately equal). But we should not introduce the index of the number of deals in the formula of strategy "goodness". This is an indicator of estimate quality (credibility of testing results). If trades are not enough for the system - the results simply cannot be trusted. I.e. it is sort of the first step of results elimination - only test/optimization results that have confidence (statistically significant) are considered. If there is trust, then it is the same for all - we can already compare "goodness" of different systems or wholesale values
People, what is the point of driving yourself into a state of frenzy from ticks in the tester if the forward cuts all efforts to zero the very next day?
No matter how much they licked the grail results for at least 10 years.
Where is the logic in that? Is it so bad that you want to drain more money? Have you nothing else to do?
They would rather spend their precious time on finding a new conceptual model.
People, what is the point of driving yourself into a state of frenzy from ticking boxes in the tester if the forward cuts all efforts to zero the very next day?
No matter how much they licked the grail results for at least 10 years.
Where is the logic in that? Is it so bad that you want to drain more money? Have you nothing else to do?
They would rather spend their precious time on finding a new conceptual model.
Let's assume that we don't need to delimit anything... just assume... :))) that we have to measure the length of a boa constrictor... not in parrots or monkeys, but in some "common" yardstick...
As I recall - this is the problem of reducing a multi-criteria problem to a single criterion . A very common problem. Thinking you have to dig here.
And how do you check whether a "new conceptual model" is the right one and makes money? :)
It should not have a distance in pips in its parameters.
Because the distance travelled in the past does not indicate a trend, but only depends on (roughly) which foot trader Vasya Ivanov got up from today.
Trying to predict the "right foot" is, in my opinion, a dense absurdity. As they call it, men are cutting wood with jigsaws.
You'd be better off spending your precious time looking for a new concept model.
Why get attached to the tester? He doesn't answer the first and most important question - model stability. Everything else is afterwards.
If a "new" concept suits you - it is econometrics, they have long and quite productively solve the question of stability of the model. Anyway, in econometrics this is the main question, and how to measure the result in terms of profitability is a tenth matter. For an unstable model any measurements are meaningless.
It should not have a distance in pips in its parameters.
Because the distance travelled in the past does not indicate a trend, but only depends on (roughly) which foot trader Vasya Ivanov got up from today.
Trying to predict the "right foot" is, in my opinion, a dense absurdity. Men cut wood with jigsaws, as they say.
so who is against it, your assumptions and preferences. But they have to be checked somehow, don't they? The question is the right test (which can be trusted)
So who is against it, your assumptions and preferences. But they must be verified somehow? The question is the right test (which can be trusted)
Only the correctness of the logic can be checked in the tester, nothing more.
The criteria for the "profitability" of the TS should be contained in its very essence and should not require statistical sampling on history.