Machine learning in trading: theory, models, practice and algo-trading - page 2565
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
A funny thing. I am picking ticks on the Hearst and I get values very different from 0.5 in the spread scale and the larger the time scale, the closer the Hearst is to 0.5. I have made a primitive system on a mask and began to substitute periods of 10, 100, 1000, 10000. All of them have approximately the same expectation. That's what an efficient market is.
I'm almost sure that you can use MO if you read about what this coefficient means.
To teach the system a pattern, depending on the value of the coefficient would not be bad
I'm pretty sure you can apply MO if you read up on what the coefficient means
it would not be bad to teach the system a pattern depending on the value of the coefficient
First we discard it, and then we combine it.
So, for each predictor we take a rule like 0.5<X<7.3, then we build a number of all possible combinations, where each rule is either included or not. We get 2^N possible variants, where N is the number of predictors. If we take all leaves of all trees obtained by boosting as initial rules, then N is the number of these leaves. In any case, even with a small number of predictors we get a large number of variants, which is connected with two problems:
1) it turns out to be a very long search enumeration
2) The method is too flexible and our compromise between bias and variance is strongly biased towards increasing variance.
In general, for small N (depends on sample size) it can work, but for large - there will be overtraining.
Come back when you are completely sure
I'm not looking for permission, you misunderstood the meaning of the post
I'm not looking for permission, you misunderstood the meaning of the post
You have to predict its dynamics, plus it's lagging
in terms of
it's just a coefficient.
you find it, make conclusions and forget about Hearst
who says in forex that the quotient has a fractal structure and plus up and down,
literally tick up, tick down.
and the fractal here is practically a triangle similar in appearance to the lognormal distribution chart, the only pattern, there is no other
which by the way was proven by Samuelson and he got a Nobel for it
so let's get the neuronics going
in terms of
it's just a coefficient.
found it, made conclusions and forgot about Hearst
who says in forex that the quotient has a fractal structure and plus up, plus down,
literally tick up, tick down.
and the fractal here is practically a triangle similar in appearance to the lognormal distribution chart, the only pattern, there is no other
which by the way was proven by Samuelson and he got a Nobel for it
So let's get the neuronics going.
At least do not shame Samuelson - he did not prove anything like that.
in terms of
it's just a coefficient.
found it, made conclusions and forgot about Hearst
who says in forex that the quotient has a fractal structure and plus up, plus down,
literally tick up, tick down.
and the fractal here is practically a triangle similar in appearance to the lognormal distribution chart, the only pattern, there is no other
which by the way was proven by Samuelson and he got a Nobel for it
So let's get the neuronics going
At least don't shame Samuelson - he didn't prove any such thing.
It would do you good to educate yourself.
What conclusions were drawn?
https://www.mql5.com/ru/forum/375928/page2
If 0 ≤ H < 0.5 - prices are fractals, the validity of FMH is proved, there are "heavy tails" in the distribution of variables, antipersistent series, i.e. negative correlation in price changes, pink noise with frequent changes of price direction;