Analyse the most important STATISTICAL characteristics of the pattern and choose a method of trading on it. - page 3
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
You can also use percentiles, it's easier to calculate, you need more data so there are no surprises...
Gave direction on where to dig.) Although, there's a lot you can do....
I'll read about percentiles, thanks )
Something tells me the odds are going to be close to 50%. :)
What is it? What's the feeling?
Probably experience. :)
Oh, how many wondrous discoveries
The spirit of enlightenment
And Experience, [son of] difficult errors,
And Genius, [paradoxes] friend,
[And chance, the god of invention]
Look up my nearest neighbour indicator in the codebase. The method is pretty simple. You set the length of the current pattern, find similar patterns from history (e.g. use correlation as distance between patterns), predict future price behaviour from past patterns by weighting their individual predictions. This is essentially the same as clustering, or RBF, or SVM, or GRNN. It all depends on how you measure the distance from the current pattern to similar past patterns. Read about GRNN and Bayes. There the prediction theory is described in terms of statistical distributions. There is a lot written about GRNN and the above mentioned prediction methods and it all boils down to one simple formula:
prediction y = SUM y[k]*exp(-d[k]/2s^2) / SUM exp(-d[k]/2s^2)
where y[k] is the k-th past pattern, d[k] is the distance from k-th pattern to the current pattern. If distances have Gaussian distribution then d[k] = (x - x[k])^2. For an arbitrary (super Gaussian) distribution, d[k] = |x - x[k]|^p, where you choose p depending on whether you want to give more weight to the closest neighbours (large p), or give all neighbours nearly the same weight (small p) as in socialism. With p=0, we have total socialism.
After getting acquainted with the nearest neighbours and GRNN, the next obvious question arises. How to measure distance between current pattern and past patterns if you take into account time axis distortions (i.e. past patterns may look like current pattern but either stretched or compressed in time). This is where the problem lies.
This is where the problem lies.
I've already eaten that dog, the question is different now. Maybe it's not quite right :)
But your publications are very interesting, thanks, I'll have a look
if the time axis distortion is taken into account (i.e. past patterns may look like the current pattern but are either stretched or compressed in time). This is where the dog is buried.
As a consequence of this statement - this dog is not discovered at the moment only due to limitations of computing resources.
This seems to be a contradiction: if you have as many computational resources as you need, any dog can be uncovered. Like, the solution to any problem depends only on the amount of computational resources available.
In general, the logic is, to put it mildly, strange. Therefore, when they say "the dog is buried there", indirectly complaining about computational insolubility at the moment, we can safely say that there is no dog there.
As a consequence of this statement - this dog is not currently unlocked only because of the limitations of the number of computational resources.
This seems to be a contradiction: if there are as many computational resources as there are, then any dog can be unlocked. Like, the solution to any problem depends only on the amount of computational resources available.
In general, the logic is, to put it mildly, strange. Therefore, when they say "the dog is buried there", indirectly complaining about computational insolubility at the moment, we can safely say that there is no dog there.
It's all done via affine transformations... and it requires minimal resources... with the right approach
It's all done through affine transformations... and it requires minimal resources... with the right approach
The grail didn't work out - the approach wasn't competent enough!
This statement has become so popular for what reason?
The grail didn't work out - the approach wasn't competent enough!
What was the reason this statement became so popular?
Well the devil is always in the details... it's not the Grail that's needed but at least something useful :)
the problem is that people don't know what they're doing, i think... and for what
prediction y = SUM y[k]*exp(-d[k]/2s^2) / SUM exp(-d[k]/2s^2)
where y[k] is the k-th past pattern, d[k] is the distance from k-th pattern to the current pattern. If distances have Gaussian distribution then d[k] = (x - x[k])^2. For an arbitrary (super Gaussian) distribution, d[k] = |x - x[k]|^p, where you choose p depending on whether you want to give more weight to the closest neighbours (large p), or give all neighbours nearly the same weight (small p) as in socialism. With p=0, we have total socialism.
After getting acquainted with the nearest neighbours and GRNN, the next obvious question arises. How to measure distance between current pattern and past patterns if you take into account time axis distortions (i.e. past patterns may look like current pattern but either stretched or compressed in time). This is where the problem lies.