Machine learning in trading: theory, models, practice and algo-trading - page 536

 
SEM:

Try adding standard deviations from Bollinger Bands or Envelopes, for channel borders, it's an interesting thing.

"Open interest from Forts", I wonder who is broadcasting the actual data on these metrics?

I don't get it again, what is "KD"?


KD is a delta cluster....

Realistically this data is broadcast from the futures market on all sorts of futures.

As for the bollinger and other indicators based on the price, I think they are secondary. Rather, they are the consequence, but not the cause of price.

Delta+Volume+OI = the basis of the instrument pricing. And the market makers have access to it. But not we do. The OM is not a universal cure, it only improves the model by 5-10%, but it can be enough.

Everybody gets this information, but very few can use it correctly. That is why it is open to all. It is one thing to know, and another thing is to be able to apply... So it goes like this .....

 

I've been thinking a lot lately about the nature of knowledge. Imagine that we have a cloze series and we start to build different indices based on this series. Standard deviations, bollinger, etc. By building on the price of the spill kind of conversions we increase the amount of knowledge in that area???? I think not. We are only expanding the dimensionality of this field by adding stochastic sub-dimensions and whatnot. YES we increase the dimensionality of the information field, but does the knowledge in this field increase??? I think not. No matter how we twist this time series, transforming it into more and more complex calculations, the knowledge we have within this field, no matter how large it may be, we do not increase. We can increase the knowledge by adding to this field completely different data taken from another time series. In other words, no matter how much you spin CLOZE, we can gain additional knowledge in this field by adding volume or delta or OI or all of them. It is the addition of new data taken from other sources that leads to an increase in knowledge in this sample of the information field.


Right now I am using Delta and Volume. But no matter how much I transform them, the knowledge can be added only by adding OM or Market Expectations (smile). Also adding knowledge to the stat sample will serve to get data from the Shanghai futures market exchange + if you add European indices or stocks + if you add moon phases, it will also increase the KNOWLEDGE in the sample. Otherwise... No matter how much you twist or convert CLOZE, the knowledge in this area or the sample will not increase. IMHO

 
Mihail Marchukajtes:

KD is a delta cluster....

This data is actually transmitted from the futures market.

As for the bollinger and other indicators, which are based on the price, I think they are secondary. Rather, they are the consequence, but not the cause of price.

Delta+Volume+OI = the basis of the instrument pricing. And the market makers have access to it. But not we have it. The OM is not a universal cure, it only improves the model by 5-10%, but it can be enough.

Everybody gets this information, but very few can use it correctly. That is why it is open to all. It is one thing to know, and another thing is to be able to apply... So it's like this.....

I meant "standard deviations" for the calculated data, not for quotes.

example on the chart (combination of bars for quotes and indicator), entry points are well traced.


 
SEM:

I meant "standard deviations" for calculated data, not for quotes.

Example on the chart (combination of bars for quotes and indicator), entry points are well traced.



It is in this view of the market. Here and now. Not the fact that in the future these settings will be relevant. But yes, I also use indices built on data other than clos kotir. It makes sense. You see what the thing is. The most important thing in the market is not the level of profit or the profitability of the system. The most important thing in the market is the stability of everything. The only thing that is stable in the market now is losing deposits. That's where they say stability is a sign of skill. But we do not need this stability. If you set up the indices now, it does not mean that they will work tomorrow. IMHO!!!!!

 
SEM:

I meant "standard deviations" for calculated data, not for quotes.

example on the chart (combination of bars for quotes and indicator), entry points are well traceable.



the squares were drawn by the ears

 
Mihail Marchukajtes:

I've been thinking a lot lately about the nature of knowledge. Imagine that we have a series of clones and we start to build different indices based on this series. Standard deviations, bollinger, etc. By building on the price of the spill kind of conversions we increase the amount of knowledge in that area???? I think not. We are only expanding the dimensionality of this field by adding stochastic sub-dimensions and whatnot. YES we increase the dimensionality of the information field, but does the knowledge in this field increase??? I think not. No matter how we twist this time series, transforming it into more and more complex calculations, the knowledge we have within this field, no matter how large it may be, we do not increase. We can increase the knowledge by adding to this field completely different data taken from another time series. In other words, no matter how much you spin CLOZE, we can gain additional knowledge in this field by adding volume or delta or OI or all of them. It is the addition of new data taken from other sources that leads to an increase in knowledge in this sample of the information field.


Right now I am using Delta and Volume. But no matter how much I transform them, the knowledge can be added only by adding OM or Market Expectations (smile). Also adding knowledge to the stat sample will serve to get data from the Shanghai futures market exchange + if you add European indices or stocks + if you add moon phases, it will also increase the KNOWLEDGE in the sample. Otherwise... No matter how much you twist or convert CLOZE, the knowledge in this area or the sample will not increase. IMHO


The cardinal rule of theanalysis is what? That's right, price takes everything into account. Nonparametric econometrics(which includes MO) is exactly the study of consequences, not causes... complicated? yes, but who would argue :)

 
Maxim Dmitrievsky:

the squares are far-fetched by the ears

Possibly. On what date and symbol can we test the hypothesis?

 
SEM:

Maybe. On what date and symbol shall we test the hypothesis?


Well I can even see from the graphics that they are not exactly drawn there... it won't work, it's too simple)

 
Maxim Dmitrievsky:

I can even see from the graphics that they are not exactly drawn there... it will not work, it's too simple)

I gave an example with areas as rectangles and candlesticks as vertical lines. And I didn't say about the great accuracy, just the regularity.

 
Maxim Dmitrievsky:

What is the main rule of tehanalysis? That's right, price takes into account everything.

It's a good rule.) It automatically follows that price takes into account not only the present, but also the foreseeable future. It automatically implies that it is impossible to forecast anything.