Market phenomena - page 41

 
Neutron:

Conclusions:

1. The cumulative sum of small increments is roughly equal to the sum of large increments. In other words, no matter where the market drifts slowly in small steps, it will always return to its original position with several large and sharp movements.

There is random wandering set by small investors and strong corrective moves orchestrated by large investors?

2. ?

OK, I will not deliberately jump to any fundamental conclusions just yet. I suggest to come back later.

If interested, here seem to me to be worthy goals:

(1) to find a more intelligible classification of the division of the "mix", i.e. the initial quotation. I.e. a more proper "tailing off"

(2) Understand the properties of the "betta" process, if it is predictable, it is cool, said - mildly

 

Conditionally, we can divide the day into two areas with high and low volatility for the selected instrument. As a hypothesis, we can assume (based on the obtained data) that the market tends to win back the price movement obtained during the period of time with low volatility. It is important here that vigorous corrections should not be randomly distributed for a day but concentrated in the area of higher volatility. Then a well-defined TS emerges.

It is necessary to listen to the opinion of colleagues...

 
Neutron:
Need to listen to the opinion of colleagues...

Why? Do you think it will have any effect on the properties of thick tails? I don't think so. The fact that they set the "rhythm" and their weight in shaping quotes is very large is already just clear after research. In fact, all big deviations of trajectories organize sequences of "fat tails". Exactly organise them, how they do it is a mystery.

PS: you are one of the peers, so give an opinion. If you think it's bullshit, no problem. I already know it's not. The question is to study it thoroughly and bring it to application. But as long as this thing is in the queue.

:о)

 
Neutron:

Conditionally, we can divide the day into two areas with high and low volatility for the selected instrument. As a hypothesis, we can assume (based on the obtained data) that the market tends to win back the price movement obtained during the period of time with low volatility. It is important here that vigorous corrections should not be randomly distributed for a day but concentrated in the area of higher volatility. Then a well-defined TS emerges.

It is necessary to listen to the opinion of colleagues...

you like to add .... in time :o))))
 
Neutron:

Conditionally, we can divide the day into two areas with high and low volatility for the selected instrument. As a hypothesis, we can assume (based on the obtained data) that the market tends to win back the price movement obtained during the period of time with low volatility. It is important here that vigorous corrections should not be randomly distributed for a day but concentrated in the area of higher volatility. Then a well-defined TS emerges.

It is necessary to listen to the opinion of colleagues...

Where is the proof that there are fat tails in your sample?
 

By the way, I remembered! Regarding volatility, there is a very good book called "Probabilistic-Statistical Methods of Volatility Decomposition of Chaotic Processes" by Korolev V.

http://www.ozon.ru/context/detail/id/6298517/

Книга посвящена всестороннему описанию вероятностных математических моделей хаотических процессов и методов их статистического анализа. Рассматривается удобный класс математических моделей стохастических хаотических процессов - подчиненные винеровские процессы (процессы броуновского движения со случайным временем). В качестве аргументации в пользу указанных моделей используется асимптотический подход, основанный на предельных теоремах для обобщенных дважды стохастических пуассоновских процессов (обобщенных процессов Кокса), которые в определенном смысле являются наилучшими математическими моделями неоднородных (и даже нестационарных) хаотических потоков на временных микромасштабах. Такой подход приводит к тому, что распределения приращений рассматриваемых процессов имеют вид сдвиг/масштабных смесей нормальных законов, и дает возможность получить не только сами формальные вероятностные модели хаотических стохастических процессов, но и в некотором смысле дать разумное теоретическое объяснение их адекватности на основе минимальных предположений о внутренней структуре изучаемых характеристик. На основе представления распределений (логарифмов) приращений процессов эволюции финансовых индексов или процессов плазменной турбулентности в виде смесей нормальных законов в книге предложена многомерная интерпретация волатильности рассматриваемых процессов. Для статистического анализа хаотических случайных процессов предложен метод скользящего разделения смесей (СРС-метод), который позволяет спонтанно разложить волатильность рассматриваемого процесса на динамический и диффузионные компоненты. Большое внимание уделено аналитическим и асимптотическим свойствам смесей нормальных распределений. Систематически рассматриваются статистические процедуры численного разделения смесей, такие как ЕМ-алгоритм и его модификации, сеточные методы разделения смесей. Обсуждаются вопросы оптимальной реализации этих методов. Рассмотрены примеры применения СРС-метода к анализу влияния информационных интервенций на финансовых рынках и к анализу данных, полученных в экспериментах с плазменной турбулентностью.

Для аспирантов, студентов и преподавателей вузов, интересующихся современным состоянием исследований в области вероятностно-статистического моделирования хаотических стохастических процессов, а также для научных работников, инженеров, специалистов в области применения методов математической и прикладной статистики к анализу характеристик финансовых рынков и плазменной турбулентности.

Keywords: generalized Cox processes, mixtures of normal distributions, subordinated Wiener processes, volatility.

just in this topic, almost (there are many interesting ideas) :o))))

 

Guys, I've read through half this thread and almost got brain haemorrhoids when entering unfamiliar, rather complicated theories.

Do you really think that you have to dig that deep to develop a profitable strategy?

 
911:

Guys, I've read through half of this thread and almost got brain haemorrhoids when entering unfamiliar, rather complicated theories.

Do you really think that you have to dig that deep to develop a profitable strategy?

You see, if we are talking about a stable business, yes. If the objective is to do something different, then you can do anything.
 
Farnsworth:

By the way, I remembered! Regarding volatility, there is a very good book called "Probabilistic-Statistical Methods of Volatility Decomposition of Chaotic Processes" by Korolev V.

http://www.ozon.ru/context/detail/id/6298517/

just in this topic, almost (there are a lot of interesting ideas) :o))))

God, I'm sick of these smart guys from universities! Ever since Soviet times, they've been clever and clever. The brightest representative is Gaidar.
 
faa1947:
God, I'm sick and tired of these smart guys from universities! Ever since Soviet times, they've been clever and clever. The brightest representative is Gaidar.

But I wonder: don't you classify your poking around with Eview as cleverness? In that sense, you're a shining example, too! Maybe even better than Gaidar - he didn't have Eview, and Hedrick and Prescott weren't Nobel laureates at the time.

;)))