Hi,
Thank-you so much for this wonderful system Aroon. I have successfully installed it and its working as of now. However, I am wondering whether I would need to update it after X amount of time. (days, weeks, months)? or do i just leave it and it doesn;t need to be updated?
In addition, I was wondering what the success rate for this system is? I just downloaded it a couple of hours ago and market is really slow right now, so not much movement. And are the alerts valid? I have received a few “Buy” alerts but there is no movement, so I did not pull the trigger. Thank-you for your input.
I agree that the bar with 0 shift may have a future-looking problem. The version of Neuro Trend that I have uses bars at shift 1 and 2. There is no 0 shift bar. Perhaps we refer to different versions? My version has no version number. inputs 15 and 16 are:
input[15] = gmThreshold((iMA(NULL,PERIOD_H1,5,0,MODE_EMA,PRICE_CLOSE,h1shift)-iMA(NULL,PERIOD_H1,5,0,MODE_EMA,PRICE_CLOSE,h1shift+1))/(Point10));
input[16] = gmThreshold((iMA(NULL,PERIOD_H1,5,0,MODE_EMA,PRICE_CLOSE,h1shift+1)-iMA(NULL,PERIOD_H1,5,0,MODE_EMA,PRICE_CLOSE,h1shift+2))/(Point10));
They do not refer to bar 0.
However, we are really discussing how many angels can fit on the head of a pin. Neuro trend is a NN that predicts the next few bars of an EMA. look at the attached image.
Two curves are digital filters, the third is neuro trend. Which do you prefer? I cannot choose either.
Although neuro trend has a small mse, that is not very useful. The mse is measured over all samples (bars). It will be very small during smooth periods, and be larger when the ema pivots. But that is precisely when accuracy is needed. Since the pivots occur less often than smooth samples, they will have small effect on the mse. In fact, a simple linear predictor will have slightly worse mse, but be just as good at the pivots. Why bother with a NN?
Hi this is an excellent tread.
I am using neurotrend on 1 hour charts EUR/USD , much better i think with the assistance of the russian digital filters.
Does any one on here have a working copy of the stuggart SNNS Simulator.
I downloaded it from here ra.cs.uni-tuebingen.de/SNNS/
and cannot get it to work. So I download a compiled version from elsewhere it ran but crashed every time!!
Although the network is still accurate I do feel it would be a good idea to train it once in a while. Any help with this would be very much appreciated.
And finally I must say I have given up using linear indicators on forex, if it wasn’t for this excellent indicator neurotrend and the digital filters. I would be trading the stock indexes, for an easier life!!! Thanks for all the input on this thread.
Yes you are right -> “This thread is as good as closed now”. Its nice to see that people still interested in this indicator and are finding it useful and keeping this thread still alive.
Just to bring to your notice, back at that time I used “JNNS” software to train the network. May be you will find it easier than SNNS.
Athough I created a couple of indicators based on SVM and NN, I never could apply them successfully for trading. May be its because I am a software engineer by profession and not a trader. I invest more time in finding good forecasting algorithms than finding how to apply them. May be I should take a course on trading (:-))
I decided to support this thread further and share what I know and remember on Neurotrend. Please dont ask me for .net data or on updates to Neurotrend. All that I used and developed during this research was shared on this thread. May be I might compile a short tutorial in future such that newbies can train the neural network by their own and use it along with their trading strategies.
i downloaded the indicator as zip file. Template is loading successfully but it is showing only color in graph not all three color lines as mentioned in post. Help me out regarding this.
I am a uni student interested in looking at NN and Forex market for my dissertation.
I have read through all the posts here and would like to no more about the type of network you use (feedthrough or recurrent) and why you chose that type of network.
Arun you say you did your masters on NN what exactly did it focus on?
thank you
From the code, it looks like a simple backprop rig, using a standard sigmoid activation. I’d recommend using fann2mql for any ANN experimentation, unless you’ve got Matlab or some other uni package set up. FANN is definitely the best documented and most robust OS code you’ll be able to use in any project.
I’m guessing he chose backprop because it’s one of the simplest methods of implementing an ANN. You’ll notice this thread has been dead a long time - no answers from contributors for around 7 months. I think you’ll find better resources at the MQL codebase and forums - there are many, many discussions on using and implementing neural networks.
Good to know that you liked the indicator. As far as I remember I used a feed forward network and used Rprop algorithm for training. I used SNNS / JNNS tool to load the training patters and train the network.
well I have spent the last couple of hours viewing this post and trying to put the indicator to work…
pure waste of time…
since the basic tool the so called JNNS doesnt seem to be so free or public as mentioned because the kernel library isnt included in the installation file I dont know why one should bother…
of course there is an older version that does after some effort and luck open a java window but dont hope for something more…
