Literature. Neural networks. Genetic algorithms. Digital signal processing. Mathematics, analysis. Statistics. - page 8

 
I think this is a little off-topic, there are books posted here, but maybe you can give me some advice on the subject
 
For example, Ostrovsky S. Neural networks for information processing.
 
As promised, download statistics. You can see what people are more interested in.
Neural networks, genetic algorithms 107
Optimization methods and algorithms ---------44
Digital Signal Processing -------------61
Mathematics and analysis ------------------------33
Statistics ------------------------------------42
Time series -------------------------------41
C++ programming ---------------------38


Finance.

archive header ----------------------------47

part no.1 --------------------------------------34

part no.2 --------------------------------------2 7

part ¹3 --------------------------------------40

part #4 --------------------------------------57

Literature on Matlab -----------------------11
Literatureon Statistica --------------------10
Maple Liter ature ------------------------1

Curiously enough, only 1 download was for Maple! The literature on networks came in first place by a significant margin.
It is also interesting that the 4th part was downloaded almost twice as much as the 2nd part, although it is impossible to unpack the archive without having all the parts at the same time.
 
goldtrader писал(а) >>

Interested in colleagues' opinions on Likhovidov's student's thesis.


I have read the thesis. A curious approach is suggested by the student: to train NS by the signals of a perfect entry indicator,
essentially a zig-zag. Has anyone tried it? Maybe it works?
Files:
diplom.rar  638 kb
 
real-trader >>:
Прочёл дипломную работу. Любопытный подход предлагает студент: обучать НС по сигналам индикатора идеального входа,
по сути зиг-зага. Никто не пробовал? Вдруг оно работает?

This is not the author's original and also futile idea. He is not the first to think of it.

There are far more interesting innovative statements in this work, which almost no NN researcher ever writes about.

 
joo писал(а) >>

This is not the author's original and also futile idea.


What is unpromising, if not a secret? Is the NS overtrained or do the patterns not play out on the OOS?

 
In general, neither.
To open a position at the top of ZZ would mean doing something paradoxical. It would mean knowing that this top is maximum/minimum compared to future tops!
There is NO information at any given time that there will be a top ZZ on that particular bar, which means it cannot be taught to the network. It is just a "point" in the flow of information.
This is why, although it is not obvious, it has not been possible before and will not be possible in the future to predict bar/bar prices ahead.
Neural networks need to be trained on likely price areas, not on specific price values.
 
joo >>:
Вообще, ни то ни другое.
Открыть позицию на вершине ZZ означает совершить нечто парадоксальное. Это означало бы знание, что эта вершина максимальна/минимальна по сравнению с будущими вершинами!
Нет НИКАКОЙ информации в каждый момент времени о том, что именно на этом баре будет вершина ZZ, а значит, этому невозможно обучить сеть. Это всего лишь "точка" в потоке информации.
Именно поэтому, хотя это и не очевидно, не удавалось раньше и не удастся в будущем прогнозировать цену на бар/бары вперед.
Нейронные сети нужно обучать на вероятные области цены, а не на конкретные её значения.

This is exactly how traders (successful ones) trade with biological networks. And since probable areas change, that's why fixed stop orders (both NN-based and classic indicator-based) do not work

 
Nice branch, by the way.
 
Optimisation methods and algorithms (added)

Beiko I.V. et al. - Methods and algorithms of the solution of optimization tasks.1983.djvu
Vukolov E.A. Statistical Analysis Fundamentals in Statistica and Excel.djvu
Kuprienko N.V. - Statistics. Methods of Distribution Analysis. Sampling - 2009.pdf
Tsirlin A.M. Optimization Methods in Irreversible Thermodynamics and Microeconomics.pdf
Sharapov V.G. Manual on Problem Solving in the Course of Variational Calculus and Methods of Optimisation. pdf