Bayesian regression - Has anyone made an EA using this algorithm? - page 12
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
...
Yusuf, I'm sorry, but you're sick and tired of shoving your 18 everywhere and on every occasion.
Gauss ISC is not a throwback, it is a classic which is better and simpler than ever before and never will be. There is nothing more stupid and dumb than dividing the mathematical methods into modern and obsolete.
Yusuf, I'm sorry, but you're sick and tired of shoving your 18 everywhere and on every occasion.
Gauss ISC is not a throwback, it is a classic which is better and simpler than ever before and never will be. There is nothing more stupid and dumb than dividing the mathematical methods into modern and obsolete.
Dimitri, thanks for the remark about incorrectness, corrected, but essentially, are there any strong objections to (18), as you put it? Like, is this regression model better than (18)? MNC applies in the presence of linear dependence, and (18), apart from the linear case, equally successfully covers the nonlinear domain, while retaining all the advantages of MNC.
Next, in order for the regression to become Bayesian, the assumption is made that the eps is distributed according to the normal law.
Please, those who are Copenhagenists, correct me if something is wrong and advise me what to do next.
Throw out the normal distribution, since it is not observed anywhere in financial instruments. And instead build a histogram of the real distribution density and approximate it.
To understand the difference, it is enough to take a look at the screenshot below. The black line represents a normal distribution and the red line represents a histogram of the real probability density function.
I.e. if we just take a triangular distribution, there will be much less error. Although it is easier to take two contiguous circles whose centres are on the same horizontal line or contiguous ellipses for more accuracy, since the sides of the triangle are clearly concave.
1. Throw out the normal distribution, as it is not observed anywhere in financial instruments. Instead, construct a histogram of the real distribution density and approximate it.
2. To understand the difference, it is enough to look at the following screenshot. The black line shows the normal distribution, and the red line shows the histogram of real volatility probability density function.
3 I.e. if you just take a triangular distribution, there will be much less error. Although it is easier to take two contiguous circles whose centres are on the same horizontal line or contiguous ellipses, since the sides of the triangle are clearly concave, for better accuracy.
1. Where to approximate? Approximate what and to what?
2.
3. The error of what?
What makes you think that? Not at all. You don't have to think about it, it's like defining the scope of the Bayesian regression.
We need to determine the features that are needed to calculate the Bayesian regression. This is the first question of how to make a square circle. This is where you may realize that the Bayesian regression does not fit in at all. But we don't care... something has to be done. Suppose that the coincidence of price values of one row and the second row (in our case the line) will correspond to the maximum likelihood. And the maximal one by one path will be 1/n (n - number of bars). Although this approach is just like drawing with a pitchfork in water. So we should invent some formula which at argument 0 gives 1/n, and at increasing argument tends to 0. Then we write down the baes formula and substitute the formula we invented earlier for the probabilities. Next we need to find the maximum of the resulting function. Probably take the derivative, equate it to zero...
The result will be almost the same as the linear regression, because the initial goal was to combine the straight line and the price series.
The assumption that data on Forex has normal distribution, and thus it is the scope of Bayesian regression is why.
Forex is a lot of brokerage companies, forex companies, kitchens - European, Chinese, Bahamian, Bermuda ... There are a lot of them. None of them dominates and does not make a decisive contribution to price formation, and neither does any player on the market. The assumption is based on the Central Limit theorem of probability theory:
"The sum of a sufficiently large number of weakly dependent random variables of approximately the same magnitude (none of them dominates, no determinant contributes to the sum) has a distribution that is close to normal."(Wikipedia)
As I understand it in relation to forex. If we collect all ticks of ALL brokerage companies in one M5 bar (millions of ticks) then ticks distribution inside the bar will be close to a normal. And the older is the timeframe, the closer it is. Each particular brokerage company has its own quotes flow that differs from the dominating global flow by the measure of deprecation of this brokerage company. This dominating flow on the chart represents a curve (certainly not a straight line!) from which no brokerage company can move far.
The assumption that forex data has a normal distribution and therefore is the scope of a Bayesian regression is why.
Forex is a lot of brokerage companies, forex companies, kitchens - European, Chinese, Bahamian, Bermuda ... There are a lot of them. None of them dominates and does not make a decisive contribution to price formation, and neither does any player on the market. The assumption is based on the Central Limit theorem of probability theory:
"The sum of a sufficiently large number of weakly dependent random variables of roughly the same magnitude (no single summand dominates, no determinant contributes to the sum) has a distribution that is close to normal."(Wikipedia)
As I understand it in relation to forex. If we collect in one M5 bar all ticks of ALL brokerage companies (millions of ticks) then distribution of ticks inside the bar will be close to a normal. And the older is the timeframe, the closer it is. Each particular brokerage company has its own quotes flow that differs from the dominating global flow by the measure of deprecation of this brokerage company. This dominating flow on the chart represents a curve (certainly not a straight line!) from which no brokerage company can move far.
So you didn't understand anything I wrote?
Throw out the normal distribution, as it is not observed anywhere in financial instruments. Instead, build a histogram of the real distribution density and approximate it.
To understand the difference, it is enough to take a look at the screenshot below. The black line shows the normal distribution and the red line shows the histogram of the real probability density function.
I.e. if we just take a triangular distribution, there will be much less error. Although it is easier to take two contiguous circles whose centres are on the same horizontal line or contiguous ellipses for better accuracy, since the sides of the triangle are clearly concave.
Yuri,
Try the Laplace distribution - bilateral exponential. In my opinion, the financial data is closest to it.
Analytical estimation of maximum likelihood parameters for Laplace:
Parameter estimation[edit]
Given Nindependent and identically distributed samplesx1,x2, ...,xN, themaximum likelihoodestimatorof μ is the samplemedian,[1]and themaximum likelihoodestimator of bis
from: https://en.wikipedia.org/wiki/Laplace_distribution
So you have not understood anything I have written?
I answered your first question. About the signs I really don't understand. To find the number of bars at which the theory works? And from this to dance? I reject it at once.
"The original purpose was to combine the straight line and the price series." - If the Bayesian regression is a straight line, then it's really no good.