Machine learning in trading: theory, models, practice and algo-trading - page 2397
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
I don't understand the difference between your idea and the constant retraining of AMO in the sliding window...
You take the last n images from the current one, sorted by time, make a prediction based on them, what is that supposed to do?
You're just stupidly retraining in a sliding window like with AMO above, what's the advantage?
The point is that the length of the sliding window should not be set by a fixed value of N, but shouldbe determined from maximizing its length, provided it does not contain outdated examples.
This is similar to what is called the change point detection problem
The idea is that the length of a sliding window should not be given by a fixed value of N, but shouldbe determined from maximizing its length, provided that it does not contain obsolete examples.
This is similar to what is called the change point detection problem
N - I mean the number of counterparts, not the window size
But this approach won't work, at least not with one-dimensional data.
Unless you take very long sections as images, but in this case "dimension curse"
N - I mean the number of examples, not the size of the window
Me too. Although at the discussed level of abstraction it is not very important what exactly is meant - calendar time, the number of bars, the number of patterns or something else. In my opinion, non-stationarity will appear with any way of representing the order of time.
Me, too.
Ah, so I'm dumb.
The non-stationarity will show up with any way of representing the order of time.
I agree...
In my opinion we're just looking in the wrong place...
In my opinion the market is not a time series, and it will always be non-stationary if you look at it as a BP.
I'm trying to establish a model on levels.
Maybe someone is interested in my model of price movement...
an example for the level for sales:
Looking for a zone of accumulation of orders to buy, this is the place where most participants believe in growth from this zone... ( Zones are automatically searched using AMO )
1) A buy zone
2) the price does not react to buying and confidently passes this zone (someone has sold to all buyers on the market and is pressing down)
3) Many buyers have closed their positions by selling, thus strengthening the downward trend, but many keep buying and pray that the price returns to their entry point (I think everyone remembers himself). Thus there is an imaginary sales area and as soon as the price approaches it buyers will aggressively close their unsuccessful purchases...
Here is such a model that can be understood, explained and everything fits together with common sense ...
If you accept this model then it becomes clear why methods of forecast for BP don't and can't work in principle with market data
How well the model will be made, how well it will work... The role of MO in the model is probably to predict the zones of limit orders...
Here are some examples on the real data, the zones are built automatically
The algorithm has only seen the first 100 prices
Maybe someone is interested in my model of price movement...
an example for the level for sales:
Looking for a zone of accumulation of orders to buy, this is the place where most participants believe in growth from this zone... ( Zones are automatically searched using AMO)
How is the beginning and end of the zone determined - different classes with a given width or what?
I set a goal to study python thoroughly, there is a book of 1700 pages, I need a week more. I will need more reserches later :)
You have to go to Tibet and sit on the rocks to read and understand 1700 pages.
That's a brave thing to do. You'll get even darker with this book. Your wife will kick you out.
I set a goal to study python thoroughly, there is a book of 1700 pages, I need a week more. I will need another week :)
book title?)
book title?)
"Learning Python" by Mark Lutz, 5th edition, 2 volumes.