neural network and inputs - page 24
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
Difficult doesn't mean impossible. Anything we can discern with our eyes can be formalised, it's a matter of principle. After all, besides "trend" and "flat", a third state "don't know" can be added, thus excluding difficult situations.
You're not going to say anything specific anyway.
Trend, flat and "don't know" is called fuzzy logic. An NS with fuzzy logic is, well, how can I put this to you? Well,in short,you can't do it.
You're not going to say anything specific anyway.
Trend, flat and "don't know" is called fuzzy logic. An NS with fuzzy logic is, well, how can I put this to you? Well, anyway, you're not gonna make it.
You don't seem to be pulling much of anything, you C-literate. You should read books instead of wikipedia. I posted the solver codes for the trend/flat class split a couple of years ago, you can find them yourself.
You don't seem to be pulling much of anything, you trooper. You should read books instead of wikipedia. I posted the solver codes for the trend/flat class split a couple of years ago, you can find them yourself.No, I won't.
Any formalization of trend-flat inevitably leads to the absence of flats as such. Always.
No, I don't.
Any formalisation of a trend-flat inevitably entails the absence of a flat as such. Always.
Any formalisation by definition means the presence of both. Look at the price movement before any news when the price may consolidate for 10 hours at one level and find a trend in that section.
In order to formalise properly, you need to set a suitable criterion for the distinction. If you don't go beyond the "the price has gone by so many points" type, it won't work. Fortunately, clever people have invented the whole science of matstatistics, the textbooks of which are so disliked by freshmen. With a whole set of criteria for separating hypotheses: choose your own if you like, and deduce your own if you like.
Any formalisation by definition means both. Look at the price movement before any news, when the price can consolidate for 10 hours at one level, and find a trend in that area.
In order to formalise properly, you need to set a suitable criterion for the distinction. If you don't go beyond the "the price has moved so many points" type, it won't work. Fortunately, clever people have invented the whole science of matstatistics, the textbooks of which are so disliked by freshmen. With a whole set of criteria for separating hypotheses: you can pick and choose your own.
again, a lot of bouffe, but again, no sense........
If you know the criteria - write them, why write about the price of potatoes at the market in Tula?
What is "price consolidation"? This is also a definition from matstatistics??????????????
Mdya.... Is there anything more specific on patterns and paternals? What was written about a few pages ago?
The last few posts have a direct bearing on what I wrote about a few pages ago. Patterns and fractals. They are the ones that help to participate in the process profitably, without predicting or guessing anything.
At the beginning they build and trade systems that are not interrelated in any way, at least to make it work, as they say, the fish don't eat fish. They shuffle indicators, throw neural nets with a million nodes into the market, trying to catch who knows what.
Then, as the knowledge about the processes in the market accumulates, the construction and selection of systems, indicators, etc. becomes more reasonable. Even if the system (pattern) has been working for ten years on the history, it does not mean anything. If the system does not correspond to the trader's perception of the processes at the given stage, it does not fit in the overall picture - it is thrown in the trash.
Systems should also be based on a system. Philosophy is needed first, and the specifics later. The opposite will not work - we will be looking for phoner "curvulines" for the rest of our lives. It is better to waste a couple of years on this philosophy and try to answer the main question: What is the process?
Without answering it, there is no way to build a good MTS. Without knowing anything about the process and what is going on in general - how can you profitably participate in it? It is impossible.
Before throwing a neural net into a puddle on the next street, you have to find out whether it has fish or water at all, maybe it has dried up long ago.
The market curve is not abstract - behind it there is money, real people and their emotions. Figuring out how these emotions behave on a chart is the main task, it's the first and foremost step.
I have already given references to the Spider and I have already given references to the Spider. I have already given references to the Spider, I couldn't be more specific. I can't write in more detail than there. Another link to the book: Soros' "The Alchemy of Finance" - it says a lot about what is going on in the market.
You know the criterion - write it down, but why write the price of potatoes at the market in Tula?
Have you heard about the method of maximum likelihood? The method of moments? Criteria of agreement? The list goes on.
What is 'price consolidating'? This is also a definition from matstatistics??????????????
Have you heard of the maximum likelihood method? The method of moments? Consensus criteria? The list goes on.
It is a concept from life, which is nevertheless perfectly formalised with the above criteria.slides, slides, slides!!!!!
I've already given links to Spider, how much more specific can I get? I can't write any more detailed than that. Another link to the book: Soros' "The Alchemy of Finance" - it says a lot about what is happening in the market.
slides, slides, slides!!!!!
almost there )))) only - splines ))) not......