Recognising images ( rhetorical theme ) - page 7

New comment
 

Let me add fuel to this idle talk...

The NS is also capable of recognising phibolevels...

 
If anyone has normal recognition of any patterns from five minutes and upwards, but has not solved the problem of exploiting them, we can agree.
 
hhohholl:

Do you need to recognise patterns? Or images? ... "The truth is out there".
Debugger:
 When discussing this topic, you don't take into account the enormous pitfalls of this direction.

I see a lot of people have been working in this field and the general conclusion is that despite their innumerable wealth very few working patterns are always obtained. Without filters there is nowhere. At the same time, TC on patterns is maniacal about fitting. BLEEP! (exclamation with tears in his eyes from the bruises covering his entire butt already)
 
Calculate the correlation coefficient between the sample and the current graph. If you find an image on the WP you can even find a database of images, if directly on the price you can find similar periods in the history.
 
gip:
If anyone has normal recognition of any patterns from five minutes and upwards, but has not solved the problem of exploiting them, we can agree.

What does "normal" mean? Can you elaborate? I do not see any difficulty in recognition
 
RomanS:

What does "normal" mean? Can you be more specific? I don't see any difficulty in recognising

I'm much more interested in what is meant by the unresolved problem of their operation
 
gpwr:

My EA recognised patterns quite accurately. And much more efficiently than with the eyes, because the eyes sometimes see what they want to see. When I ran it carefully through the history, the statistics of predicting the direction was close to 50%. And profits, if there were any, were intermittent, i.e. 2-3 months of successful trades were followed by 2-3 months of complete loss.

I don't want to argue with anyone. If you have a successful EA trading by patterns, you should post the statements. Then we will have a substantive discussion.


Any pattern works much more effectively if you add some conditions to it (non-indicator ones). I also tried a lot of patterns and the result was 50/50 and I obtained very good results practically on all major pairs, the only optimization parameter was a zigzag step and the values vary from 16-20 on different pairs, the variation is not that big
 
Mischek:

I'm much more interested in what was meant by the unsolved problem of their operation

))) so one by one please.... me first
 
RomanS:

)))) so one by one please.... me first


It seems that gip promised something, so go for it ))

 

If there is a pattern recogniser but no profit, give it to me. If I manage to profit from your pattern recognition algorithm with my expert system, then we can agree on sharing. Depending on the results. You can give me not the algorithm itself, but the primary result of its work to try out. Need: The number of recognized situations at least about once a day.

---

Actually a successful pattern recognition algorithm is rare. And there are a lot of bad ones.

Candlestick patterns working with 1-8 candlesticks are not needed. We do not need patterns shorter than 15 minutes. We need informative, meaningful patterns and situations. I make a building out of these bricks.

123456789101112131415
New comment
Reason: