How to find patterns in the stock market - Articles, Library comments

Реter Konow 2019.12.07 13:36 #111

In discussing market movements, we are addressing the presence/absence of patterns in a narrow sample of data, not realising that even if they exist in a movement, it does not mean that the data will allow us to extract them. And no amount of maths will allow us to solve an equation where there are more unknowns than known. It turns out that we accept in advance the condition to look for a pattern in the data space, in which its trace may be so blurred that no analysis will reveal it. But it is there and it can be seen in other data that we do not have. What to do? For starters, realise that the Market carries a greater variety of data than we have, and if we combine all types of data, patterns will emerge more strongly. In short, we need to not only look for patterns, but also enrich the diversity of market data.

Discussion of article "Horizontal Discussion of article "Developing Discussion of article "Custom

Renat Akhtyamov 2019.12.07 13:38 #112

Maxim Dmitrievsky:
Just recognise that I've found a useful pattern, the rest is philosophising. But you can go on like, Stirlitz stood his ground, it was Mueller's favourite torture.

If you have that belief, you have to take the period from the beginning of history and break it down to the present day.

If a pattern can be found, it will definitely work.

We are waiting

Discussion of article "Visual Discussion of article "LifeHack Discussion of article "Studying

Maxim Dmitrievsky 2019.12.07 13:43 #113

Renat Akhtyamov:

waiting

wait

Renat Akhtyamov 2019.12.07 13:46 #114

Maxim Dmitrievsky:

wait

What do you mean?

a bug is found, everything's gone, and it's all rubbish?

Реter Konow 2019.12.07 13:47 #115

For example, if we had data on the open interest and capital volumes of trading participants, we would put them into our overall statistical study and see much more. The regularities would be immediately apparent. But no. We look where they let us look. And the results are corresponding.

Discussion of article "Custom Discussion of article "Practical Discussion of article "Color

Maxim Dmitrievsky 2019.12.08 14:46 #116

fxsaber:

It's a chicken and egg question. You can convince yourself of the correctness of any approach.

From my point of view, you have done implicit optimisation. Any study is implicit optimisation, which is always a subset of explicit optimisation.

Non-optimisation is the absence of a statistical study. Roughly speaking, when you made a hypothesis without data and it was confirmed.

As for the candy, we are dealing with the most primitive TS on the MA. The optimiser will pick up the result, if there is one, much better than classical studies.

The only difference is the use of a time filter. The very thing that has been used by the night visionaries for many years.

Honestly, I don't understand why this shit happens.

Some will say it's due to the moustache drawers. But that's the chicken and egg dilemma again.

Fact is, the dumbest TC shows results that are mind boggling. Discouraged and want to find the catch.

I reread your arguments once again, it is not clear what they are aimed at, and did not catch the point.

In fact, we have: a regularity has been found with the help of boxplots, it has been confirmed by the TS test.

It is shown that the pattern is weaker on another interval, so the TS does not work there(with the original parameters).

You optimised the TS and saw that it is possible to pull the TS to + on that interval, which I did not deny, I only showed that there is no such a bright pattern there. It should not be excluded that different brokerage centres have different quotes and the results may differ.

Any arguments from you and other opponents about what it is:

just MAshka and overoptimisation
The pattern was not found through boxplots.
you would have easily found it yourself through optimisation (without knowing where to look at all).
it's not a pattern.

Doesn't stand up to any criticism, just nagging due to some specific misunderstanding of the material.

Because of such comments, readers may get the impression that the article is yuck, although this is absolutely not the case. Which was confirmed by subsequent comments from less "savvy" people, who simply began to echo your words without understanding the meaning of what was said.

H.Y. with such sketches you can confuse anyone and devalue the approach

Discussion of article "The Discussion of article "Horizontal Discussion of article "Reversing:

Stanislav Korotky 2019.12.08 19:27 #117

Explain for the nerds how exactly the moustache was built. In the above code, for example:

Monthly_Returns.boxplot(column='close', by='month', figsize=(15, 8))

which I understand means the default setting is 1.5 IQR, and the whiskers are symmetrical.

And further down in the text:

Усы ящиков дополняют распределение, охватывая 99% дисперсии всей выборки

Is there a link to the formula or documentation?

Discussion of article "Developing Discussion of article "How Discussion of article "The

Maxim Dmitrievsky 2019.12.08 19:50 #118

Stanislav Korotky:

Explain for the nerds how exactly the moustache was built. In the above code, for example:

which I understand means the default setting is 1.5 IQR, and the whiskers are symmetrical.

And further in the text:

Is there a link to the formula or documentation?

The moustache boxes are always built the same way, depending on the distribution. The parameters passed in are the closing prices and the period for the guppy, in this case monthly. Next is just figsize

In Russian in Wikipedia, I think, normally written, compared to pdf

Whiskers are symmetric with symmetric distribution, respectively

https://ru.wikipedia.org/wiki/%D0%AF%D1%89%D0%B8%D0%BA_%D1%81_%D1%83%D1%81%D0%B0%D0%BC%D0%B8

Discussion of article "Gap Relative Vigor Index - Relative Vigor Index -

Реter Konow 2019.12.08 20:02 #119

Objectivity for the sake of objectivity - no pattern can be proven by statistical study. Statistics is not used to prove hypotheses or theories that we put forward. Statistics can "back up" a theory of the relationship between some cause and effect, and allow for further speculation, but it doesn't prove anything. A regularity is proven by the 100% fact of a cause and effect relationship. Using statistics as a proof base is like proving Pythagoras' theorem not by formulae, but by millions of measurements of ratios of isosceles triangle sides.

Discussion of article "LifeHack Discussion of article "Practical Discussion of article "The

Maxim Dmitrievsky 2019.12.08 20:13 #120

Реter Konow:

Objectivity for the sake of objectivity - no pattern can be proven by statistical study. Statistics is not used to prove hypotheses or theories that we put forward. Statistics can "back up" a theory of the relationship between some cause and effect, and allow for further speculation, but it doesn't prove anything. A regularity is proven by the 100% fact of a cause and effect relationship. Using statistics as a proof base is like proving Pythagoras' theorem not by formulae, but by millions of measurements of ratios of isosceles triangle sides.

and neural networks are like pyramids.

A pattern is a set of repeated events cause -> effect, supported by statistics and experiment. The greater the repeatability, the more statistically significant the conclusions about its presence. A regularity can be local, on some piece of the graph, or global.

It's time to stop engaging in demagoguery with demagogues. But there is nothing more to talk about on the forum.

Why is no one Discussion of article "Applying Discussion of article "The

Discussion of article "Exploring Seasonal Patterns of Financial Time Series with Boxplot" - page 12