From theory to practice - page 417
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
Here's what I'm thinking.
If the distribution of a sample of, say, 1,000,000 ticks is unstable (and I still can't get that volume on my exponential time) and changes variance over time, then it turns out that in my case neither the arithmetic mean nor the weighted average can be used as a measure of central tendency.
This leaves me with the median.
Channels must be plotted in relation to the median. Is that right?
This instability may well be a consequence of non-stationarity (not necessarily, but most likely). In the case of non-stationarity, any sampling quantity (moments, quantiles, etc.) is likely to be meaningless. I didn't write about the fundamentals of theorizing for nothing - sampling quantities are usually counted for a series of equally distributed random variables. In the case of non-stationary increments, they are differently distributed (by definition).
Guys, you need to take an inverted autocorrelation (regression) and use it to build increments, watch errors, distributions, teach NS or whatever you like
The only difference between the effective and fractal market theories is that the series is not autoregressive but inverted. Hence there is a memory, such a series is predictable.
Moreover, it may be an inverted autoregressive of n-th order, even the Erlang thinning may be suitable here.
and chop dough by the ton.
I will finish the indicator when I will come back from holiday. But do it yourself, don't be lazy. It is not written anywhere in books about it, so there is a chance :)
Guys, you need to take an inverted autocorrelation (regression) and use it to build increments, watch errors, distributions, teach NS or whatever you like
The only difference between the effective and fractal market theories is that the series is not autoregressive but inverted. Hence there is a memory, such a series is predictable.
Moreover, it may be an inverted autoregressive of n-th order, even the Erlang thinning may work here.
and chop dough by the ton.
I will finish the indicator when I will come back from holiday. But do it yourself, don't be lazy. It's not written anywhere in books, so there's a chance :)
Do you mean Inverse Autoregressive Flow (IAF)?
Do you mean Inverse Autoregressive Flow (IAF)?
I'm sorry, I don't know the name, maybe
need to read, if there sample is divided into 2 equal parts, the first is reversed mirrorwise and counted as an AF or autoregressive by values (lag value is taken from the 2nd sample), then yes
and window size should change when searching for least error on samples, i.e. take 4 points, divide by 2, flip 2nd piece mirrorwise, count correlation, take 6 points, then 8, etc. The bigger the window and the higher the correlation, the more interesting for trading
Unfortunately, I don't know the name, maybe
it is necessary to read, if there a sample is divided into 2 equal parts, the first is reversed mirrorwise and it is counted as an akf or autoreg. by values (lag value is taken from the 2nd sample), then yes
and window size should change when searching for smallest error on samples, i.e. take 4 points, divide by 2, flip 2nd piece mirrorwise, count correlation, take 6 points, then 8, etc. The bigger the window and the higher the correlation, the more interesting for trading
Are you delusional?
Meaning?
Unfortunately, I don't know the name, maybe
I need to read it, if it divides sample into 2 parts, the first one is reversed mirrorwise and counts as an acf or autoreg. then yes.
Apparently it's something else, but also from the field of neural networks.
Still, I don't think there are any ways to reduce the price series to some kind of stationary process. Rather, one should adapt the methods available for non-stationary processes (e.g. the decay problem)
Also, the sampling ACF (like the sampling distribution, moments etc.) only makes sense for a stationary process. In case of non-stationary process there will be problems like TCApparently it's something else, but also from the field of neural networks.
Still, I don't think there are any ways to reduce the price series to any stationary process. Rather, it is necessary to adapt the methods available for non-stationary processes (e.g. the decay problem).
Probably not the entire series, but separate parts of it may be reduced to such a state using this method with elimination of "bad" ones.
But it is easier to finish later and see than to explain in your own invented terms :)
Besides, the selective ACF (as well as the selective distribution, moments, etc.) is meaningful only for a stationary process. In the case of non-stationarity there will be problems like with TC
The search for a stationary process takes place by cointegrating the chart with itself, but with an inverted part of it. Unsuccessful parts are skipped and no trading takes place
But I'm tired of inventing new entities :) then I will demonstrate it on the indicator, first of all to myself
another perversion of the price
How would one simply explain the utopia of such an occupation...?
Ah, oh!
Suppose I go to a shop and all of a sudden start figuring out how much cheaper or more expensive the price has become? // Or even worse - pulling a Fibo on the price tag.
I state - I've made an estimate of the past.
I'm not likely to have a prediction from this analysis, am I?