Request for developing Advanced Neural mt4 indicator | Page 2 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

Neural Networks: Myths And Reality 19 replies

neural network based systems in forex 22 replies

Neural Network anyone? 12 replies

The "Better" neural network 53 replies

Neural Programming for Profitable trading 22 replies

Attachments: Request for developing Advanced Neural mt4 indicator
Exit Attachments

Request for developing Advanced Neural mt4 indicator

  • #21
  • Edited 4:57am Sep 29, 2016 4:12am | Edited 4:57am
I finally had the time to put the indicator together, which you can find attached.


Remarks:
- I did not really understood, what the "gap velocity" is
- It only works on the M1 chart.
- The Analyzation usually ends at around 30000 bars, which still means 6000 analyzed M5 candles. (To be completely honest, I do not really know the reason why it does not go further.)
- Due to slower machines, which might get throttled down by displaying 12.000 text object, I have built in a security switch. At home, I didn't have a problem with it, but a small, low-end laptop I rarely use to check things suffered greatly from it.
- Since at the moment, there is a direct comparison between the last detected M5 candle and the previous candles, even with refining the velocity numbers to a two decimal number, the result at the place of the prediction will be more likely the text: "No matching bar found". A different logic might get a result, but it might harm the accuracy of the prediction. (If we truly believe, that history will repeat itself in a M5 timeframe.)

Gorraf
Attached File(s)
File Type: ex4 ProblemSolving_Xbeast_160927_ind.ex4   17 KB | 263 downloads
  • #22
  • Sep 29, 2016 6:38am Sep 29, 2016 6:38am
Quoting Gorraf
I finally had the time to put the indicator together, which you can find attached. Remarks: - I did not really understood, what the "gap velocity" is - It only works on the M1 chart. - The Analyzation usually ends at around 30000 bars, which still means 6000 analyzed M5 candles. (To be completely honest, I do not really know the reason why it does not go further.) - Due to slower machines, which might get throttled down by displaying 12.000 text object, I have built in a security switch. At home, I didn't have a problem with it, but a small, low-end...
Really great work friend and I found some matches also but it is not predicting the candle direction instead it is showing no. of 1m bull and bear bars and their velocity.As you mentioned,it is not making many predictions.
I have one idea you can make this small change hopefully.

Make the decimal point of velocity in 1 digit and consider sequence of m1 candles or pattern ( ex - bull bull bear bull bear).

Gap velocity is the rate of change of gap by time. (gap is the distance between previous candle close and present candle open) and consider it's sequence or pattern also for 5 m1 candles. Consider the gap velocity in 1 digit also.

hopefully by this adjustments this system will predict every m5 candles.

congrats gorraf and really thank you for your work.
  • #23
  • Sep 29, 2016 8:38am Sep 29, 2016 8:38am
Quoting Xbeast
{quote} Really great work friend and I found some matches also but it is not predicting the candle direction instead it is showing no. of 1m bull and bear bars and their velocity.As you mentioned,it is not making many predictions. I have one idea you can make this small change hopefully. Make the decimal point of velocity in 1 digit and consider sequence of m1 candles or pattern ( ex - bull bull bear bull bear). Gap velocity is the rate of change of gap by time. (gap is the distance between previous candle close and present candle open) and consider...
- digits changed
- pattern added (to the prediction part as well)
- gap velocity added
-

Gorraf
Attached File(s)
File Type: ex4 ProblemSolving_Xbeast_160927_ind.ex4   23 KB | 227 downloads
  • #24
  • Sep 29, 2016 9:10am Sep 29, 2016 9:10am
Quoting Gorraf
{quote} - digits changed - pattern added (to the prediction part as well) - gap velocity added - Gorraf {file}
buddy you're super fast.
No signal for 1/2hr
what might be the problem ?? any idea??
  • #25
  • Sep 29, 2016 9:45am Sep 29, 2016 9:45am
Quoting Xbeast
{quote} buddy you're super fast. No signal for 1/2hr what might be the problem ?? any idea??
There is 6 criteria which a new 5M bar should met
- Bullish bars
- bearish bars
- bullis velocity
- bearish velocity
- pattern
- gap velocity

I think that, even if there is around 6000 bars to be sampled from, the chances of finding a match is pretty low.
I would leave out gap, velocity all together, and set some kind of alert to either bullish or bearish velocity, because, the time when you want to act is usually at volatile markets. When you have five up candles before you, it suddenly gets way more interesting what has been happened in a similar situation.

(But, that would be a quite different indicator with a quite different mindset. )
  • #26
  • Sep 29, 2016 10:02am Sep 29, 2016 10:02am
Quoting Gorraf
{quote} There is 6 criteria which a new 5M bar should met - Bullish bars - bearish bars - bullis velocity - bearish velocity - pattern - gap velocity I think that, even if there is around 6000 bars to be sampled from, the chances of finding a match is pretty low. I would leave out gap, velocity all together, and set some kind of alert to either bullish or bearish velocity, because, the time when you want to act is usually at volatile markets. When you have five up candles before you, it suddenly gets way more interesting what has been happened in...
yes you're correct but we need to wait till volatile condition 5Up OR 5Dn to get a single signal.

1.5 bar pattern = repeats often
2.gap velocity (since it's 1 digit) =repeats often
3.bull&bear velocity (since it has decimal point)= repeats rare

so,if we remove bull and bear velocity is there any chance to predict every candle??
  • #27
  • Sep 29, 2016 10:28am Sep 29, 2016 10:28am
Quoting Xbeast
I could see lot of programmers out here in forexfaxtory.So I'm glad to share my idea. I'm hoping to develop simple price action neural Indicator which is not like traditional one that shows signals.I'm looking for an indicator that labels every candle and makes projection. Steps: 1.calculate number of bull actions(bull ticks) per candlestick and display it on top of it in blue color. 2.calculate number of bear actions(bear ticks) per candlestick and display it on bottom of it in pink color. 3.correlate the blue and pink numbers with historical candles....
Nice for sharing your idea. May we know how to apply this indicator (if this indicator already available) in real trading ? thanks for your information
  • #28
  • Sep 29, 2016 10:44am Sep 29, 2016 10:44am
Quoting ronald_fsm
{quote} Nice for sharing your idea. May we know how to apply this indicator (if this indicator already available) in real trading ? thanks for your information
We are building a system to predict future based on historical data,we believe that the market pair movements are the character of each currency involved in it.So we hope that predicting their movement early we can trade on remaining part.
Indicator is not yet ready and we are figuring out.
We welcome innovative ideas.
  • #29
  • Oct 4, 2016 8:57am Oct 4, 2016 8:57am
Quoting Xbeast
{quote} yes you're correct but we need to wait till volatile condition 5Up OR 5Dn to get a single signal. 1.5 bar pattern = repeats often 2.gap velocity (since it's 1 digit) =repeats often 3.bull&bear velocity (since it has decimal point)= repeats rare so,if we remove bull and bear velocity is there any chance to predict every candle??
The answer to the question is both yes and no!

Let's break this all into pieces.
The aim of the EA is to have positive detections. Even if not in hundred percent, but if the detection rate would be under 80%, then the EA would cause more losses, than profits.

(Actually, we would only need around 66,6% positive predictions, but since false predictions would cause more diverse samples, which would lead to more diverse erorrs in the predictions, which would lead to losses, I guess the correct number is around 80%.)

Now, while we aim for winning money, and assume, that we can correctly predict 80% of the candles based on past data, we could say, that the chart is more or less, a not perfect loop or a circle.

While this is not the chase, the theory, in its current is proven wrong.


So, the answer is yes, because the indicator will find matches more frequently if we have less restrictions, but the predictions won't be neessarily true, which would make the indicator pointless and this is why the answer is more like a no.


However, I am not here to do any kind of intellectual reasoning, I am just a code junkie, so if you want I can still modify the EA and you can prove me wrong.

Gorraf
  • #30
  • Nov 22, 2016 10:04pm Nov 22, 2016 10:04pm
Inserted Video
  • #31
  • Aug 29, 2017 1:14am Aug 29, 2017 1:14am
The most advanced FREE neural network & astronomical trading signals http://www.myfxpair.com/astrological-trading-signal/ which can be used for forex ,binary options and gold,silver trading.
  • #32
  • Last Post: Edited 10:20am Apr 10, 2019 9:26am | Edited 10:20am
A
0 traders viewing now