Recognising images ( rhetorical theme ) - page 3
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
Diamant mentioned the toolkit from Alpari called Autochartist. Also about pattern recognition, there is a solution from UV. ForexTools, i.e. the tool ft.Dejavu. But it does not recognise technical patterns, but searches the history n-bars backward for the pattern, which is forming on the chart at the moment. Both Autochartist and ft.Dejavu - both must be based on correlation of the reference image with a fragment of the chart. At least, that's what I would do. You could also search online for FineReader technologies - but that would be a clear overkill
finereader is geared to the recognition of the printed alphabet, the subject of recognition of printed text has already been solved almost perfectly
Neuronets, which have been honed for this purpose, cope with it ideally...
There are even freely available NEUROSET codes customized for the alphabet.
Good afternoon!
Since I am writing to order, interesting letters are coming in
I won't give any examples from the letters... just summarizing the experience of communication over a long period of time
the gist is about the following:
When a double top is formed, "sell" - "buy" depending on whether the top is bull or bear
--
The topic itself is interesting, of course any image can be transferred to digital
then by averaging out the image... for example to distinguish between a cat or a dog in the picture :-)
( I remember reading somewhere - a computer cannot trade unless it learns to distinguish a cat from a dog)
When I say to a surprised customer that a program that can tell the difference between a dog and a cat will cost a lot of money
The correspondence on the terms of reference usually stops...
I don't think they care how to recognize...
because judging by the mission statement: TO SELL AT A DOUBLE TOP...
so there is no DESCRIPTION in the order how to recognize it, what a double top is...
here the customer believes that I myself will figure out how to recognize a double top
The TOR should be specific. That is, the customer himself should specify what a double ver tex is. At the very least, give a specific example of a double vertex and all other shapes that are not a double vertex. A concrete example can be a section of history data for some symbol, i.e. from the bar with such and such time to the bar with such and such time we have such a figure.
In this case we need to make a block that will distinguish double tops from any other pattern. And further we form a trading signal according to the recognition results.
Simply put, we should separate flies from cutlets, make two blocks: pattern recognition, i.e. classification of patterns and trade signals forming according to the results of this recognition.
Neural networks are quite suitable for classification tasks.
Maybe look towards exotics: Renko, tic-tac-toe, ...
Maybe look towards exotics: Renko, tic-tac-toe, ...
No. You have to do this once a day at the 13th second of the 8th minute of the 5th hour.
Gathering statistics should not only focus on the repeatability of figures but also on their scale
The market is changing ... let's take 2007 - the Euro was moving from a high to a low of 60 pips on average.
2009-2010, the Euro was averaging 120p.
it is the same with the figures - they should be scaled up at least by the average move in the last sector
Long time ago (a year or two ago, I even wrote here) I decided to play with it. Just raised a similar issue.
Decided to do so - threw on ... ahem ... ... I plotted a "grid" on a chart (Dash, Dash, RSI, Bollinger, CCI). The cell n-bars by m-pips. The cell gets "graph" - 1, empty - 0.
Then, vertically (I decided not to bother with arrays), I encoded the resulting zeros-units into letters. Correspondingly columns were coded into words.
It turned out that every closed bar was a word and the "teacher", looking into the future, said it was worth buying/selling/on the fence.
After that I tried to analyse it all, by coarsening the digitisation (by enlarging the cells), trying to scale it (for example, I got the word "head-shoulders" with 12 pips). ...
In the end ... pseudoscience won :)
Sometimes I dabble in this way: I write the code of a "recognizer" based on ZZ or fractals, then I put it into a single informer, which looks through several currencies in several TFs, and gives messages if it finds anything.
After that I perform all-round analysis and decide whether I should enter this subject or screw this image...
a couple of years ago Victor and I did http://xrust.ucoz.net/forum/8-36-2
a couple of years ago Victor and I did http://xrust.ucoz.net/forum/8-36-2