From theory to practice - page 79

[Vladimir]

Alexander_K2:

I have, for example, compiled what I call technical passports of probability densities of currency pair increments.

These datasheets do not describe the pairs, but the filter settings. Try to check how they change, especially during the summer holidays.

[[Deleted]]

Vladimir:

Thank you.

But... I really don't know how to graph SB. Not only that, I don't trust any of the pseudo-random number generators at all until I check them myself. And that's an extremely hard thing to do, so I never compare with graphs.

We're not going to flip a coin 20,000 times to generate a series ourselves.)

[Roman Shiredchenko]

Vladimir:

It's a whole branch, I've been tortured to find it. Is there really a standard defining what "SB" is?

Yes. There is no standard. There are variations.

[Vladimir]

Максим Дмитриев:

is to move a point by a distance +1 or -1 with probability p. Where (0<p<1).

In the case of a coin p is always strictly 1/2.

Forced to repeat the question, "This is a whole branch, tortured to find. Is there really a standard that defines what "SB" is?

I was interested in the availability of a standard. GOST, ROST, CEB, IEEE, finally. Is there a standard?

P.S. While I was repeating the question, the answer was that there are no standards, the link does not lead to them. This is the case. Was this answer correct? I don't know.

The standards that are supposed to account for variance instability, I've gathered, but haven't read it yet

GOST_R_54500_1-2011.pdf
GOST_R_54500_3-2011.pdf
GOST_R_54500_3_1-2011.pdf
GOST_R_54500_3_2-2013.pdf

I do not answer, but at first glance they do not contain the words "accidental wandering".

[Mykola Demko]

Vladimir:
And the most complex?

And the most complicated one is modelling the market stack with the order matching algorithm and the market maker algorithm, only on incoming flows of buy and sell market orders you feed PRNG.

[[Deleted]]

Alexander_K2:

only immediately rushed to price deviations from the average, not increments. And the process of the price itself, in contrast to the increments, has become non-stationary! It is more difficult to analyse. Give it up. Analyze ONLY the increments. It will give you a great deal.

How can you trade increments? At least you can trade deviations.

Alexander_K2:

I have, for example, compiled what I call technical passports of probability densities of currency pair increments. It makes sense because the process of increments is stationary.

Where did you see that increments are tested for stationarity.
Stationarity is usually checked for a process.
deviations from the mean.

[Vladimir]

Максим Дмитриев:
We're not going to flip a coin 20,000 times to generate a row ourselves.)

Why generate? There aren't enough archive sources. I do not understand what is the purpose of identifying the type of distribution. What good is the Kolmogorov-Pearson criterion to give so much (a) and so much (b) we need for 97% confidence that we have not rejected a false hypothesis. When estimating by method of moments (if you estimate by method of maximum likelihood or Bayes, a and b will be different). Who needs the whole distribution? Even Alexander_K said that he cares about tails first and foremost. Coin flips, by the way, have also been done up to 50,000 times by at least a dozen different scientists.

[Vladimir]

Nikolay Demko:

And the most complex you simulate a market stack with an order-matching algorithm, and a market maker algorithm, only on the incoming streams of buy and sell market orders you feed PRNG.

And this would be a random walk model? Exactly SB with double logarithm in the denominator under the square root?

[[Deleted]]

Vladimir:

Forced to repeat the question, "It's a whole thread, tortured to find. Is there really a standard defining what "SB" is?

I was interested in the availability of a standard. GOST, ROST, CEB, IEEE, finally. Is there a standard?

P.S. While I was repeating the question, the answer was that there are no standards, the link does not lead to them. This is the case. Was this answer correct? I don't know.

The standards, which are supposed to take into account variance instability, have collected, but haven't read it yet

GOST_R_54500_1-2011.pdf
GOST_R_54500_3-2011.pdf
GOST_R_54500_3_1-2011.pdf
GOST_R_54500_3_2-2013.pdf

I do not answer, but at first glance they do not contain the words "accidental wandering".

https://dic.academic.ru/dic.nsf/enc_mathematics/5137/СЛУЧАЙНОЕ

[[Deleted]]

Alexander_K2:
Yes, you can! Just trade not increments, but ON increments. Many people use it, it's called "pipsing". But they are ruined by spread and commission. I understood it at once, so I gave up this method.

I don't know why I gave up this method.

---

I already told you, what will it give you if you know that the next most probable gain is 1 pip.