From theory to practice - page 892

New comment
 
Evgeniy Chumakov:


At least 1,000 times.

In stochastic process models, the so-called "mean" is a measure of the central tendency of the process. The dispersion formulas for both Gaussian process and Laplace process assume that a certain distribution of linear deviations must be formed in relation to this measure.

And if you're using a formula to calculate the variance for a Gaussian process, be kind enough to choose a measure around which a normal distribution is formed.

You can't use variance and mean separately as you wish - they are all interrelated.

P.S. I don't want to write anything, so that idiots like Tarabanov don't read it, but I have to...

 
Alexander_K2:

In stochastic process models, the so-called "mean" is a measure of the central tendency of the process. The dispersion formulas for both Gaussian process and Laplace process assume that a certain distribution of linear deviations must be formed in relation to this measure.

And if you're using a formula to calculate the variance for a Gaussian process, be kind enough to choose a measure around which a normal distribution is formed.

You can't use variance and mean separately as you wish - they are all interrelated.

P.S. I don't want to write anything, so that fools like Tarabanov don't read it, but I have to...


I haven't even calculated the variance yet.

 
Evgeniy Chumakov:


I haven't even calculated the variance yet.

:))) Remember, we were agonizing with quantiles when calculating dispersion - here, it would be nice to know exactly what the distribution is now, so that the quantile is automatically calculated. In fact, we were working with SMA at that time and within those limits we always get "heavy" tails and deposit drains. Moreover, this problem, alas, is unsolvable or very hard to solve.

So - SMA down the drain! It is good only under a constant normal distribution, but in fact - we need an average around which there would always be an asymptotic approximation to the normal. In the limit, so to speak. Only in this case you can use the formula for variance calculation, which you and I were tinkering with, with regular quantiles.

 
Alexander_K2:


So - SMA down the drain!


This has been understood for a long time.


Alexander_K2:


we need an average around which there would always be an asymptotic approximation to normal.


And I rather guessed how you did it, at least I don't see any other solution.
 
Alexander_K2:

:))) Well, sort of, yeah - I'll include it a bit later.

Actually, the contest is a funny thing. Curious to see how some people manage to make 100 trades in just 6 hours of trading or lose half of their deposit :)))

I think it's not 100 trades in loss that has become a concern for you, but 100 trades in profit.

And the collective farmers have become a considerable distance ahead of you physicists.))))

 
Uladzimir Izerski:

I think it's not 100 trades in loss, but 100 trades in profit that you've become concerned about.

And the collective farmers have become a considerable distance ahead of you physicists.))))

Nah, those who have managed 50-100 trades each are in deepest pardon. The ones with 1-2 deals are ahead.

Me and Automat are on the sidelines. We're waiting for the right moment to get out in the open.

 
Shit... Alexander wanted to write such a joke about his coming to the forefront)))) But he deletes messages more often than he writes them))) Quantum Physicist))))))))))))))
 
Evgeniy Chumakov:


In general I think that if speed (whether in the construction of the mean or variance) is not considered in this model, then no amount of kurtosis and asymmetry will help.

Although according to the tests speed is not always helpful. At the same time, all entries into trades with kurtosis < 1 and asymmetry +-1.

The only thing that will probably help is the correct average. Where to get it.

If you sit in minutes, maybe you just can't see the forest for the trees).

 
vladevgeniy:

:))) FelixWhite, darling! Write, why don't you?

Since you started to sprinkle posts on the case - I withdraw all my claims to you and apologize, if I offended you in any way.

You believe that we'll find the Grail here, in spite of all the half-drunken Tarabans, don't you?

 
OK, tempted many times, but one will have to write))))) If the real FelixWhite wrote here, I'd learn this thread by heart))))))))))) And the likes of Neo and Ataman (well heaven forbid). You could have GaryTrader ppl))) If, Alexander, you like calling me that, I'm flattered))) But, in this case, the content does not match the label)))))))
1...885886887888889890891892893894895896897898899...1981
New comment
Reason: