Metaneural EA 2014 (By metaneural)
| Gain : | +147.95% |
| Drawdown | 53.65% |
| Pips: | 1046.5 |
| Trades | 259 |
| Won: |
|
| Lost: |
|
| Type: | Real |
| Leverage: | 1:200 |
| Trading: | Automated |
Metaneural EA 2014 Discussion
Oct 15, 2014 at 16:15
916 Views
14 Replies
Member Since Feb 10, 2013
8 posts
Jan 06, 2015 at 14:52
We use various indicators to gather data in different formats but ultimately they are run through neural network processing and are utilized as patterns.
Jan 06, 2015 at 18:22
Does your EA performance based on how much indicators did you used? or like expert said, simple way (just several indicators) is the best? Well, I bet you're not using Moving Averages only just like in your video in your site.
Logic creates fortune..
Member Since Feb 10, 2013
8 posts
Jan 06, 2015 at 19:47
Hello Reximus,
No, the performance of the system is not based simply on how many indicators are used but on how the market data is processed and extrapolated into the future to determine the next probable step in the pattern of the chart. There is more information about how the EA works on our website metaneural.com
Thank you for your interest.
Kind Regards,
No, the performance of the system is not based simply on how many indicators are used but on how the market data is processed and extrapolated into the future to determine the next probable step in the pattern of the chart. There is more information about how the EA works on our website metaneural.com
Thank you for your interest.
Kind Regards,
Member Since Feb 09, 2010
226 posts
Jan 06, 2015 at 21:50
Is your EA broker senstive? I ask because some of your gains are very low in terms of pips.
Member Since Feb 10, 2013
8 posts
Jan 07, 2015 at 08:05
Hello Fxtrader10,
We recommend using an ECN broker because they provide smaller spreads, however the EA can be set to run on non-ECN brokers as well. Thanks for joining the discussion. Cheers.
We recommend using an ECN broker because they provide smaller spreads, however the EA can be set to run on non-ECN brokers as well. Thanks for joining the discussion. Cheers.
Jan 07, 2015 at 08:41
(edited Jan 07, 2015 at 08:47)
No, the performance of the system is not based simply on how many indicators are used but on how the market data is processed and extrapolated into the future to determine the next probable step in the pattern of the chart.
Your site suggest we gather the data on one currency pair and on one specific timeframe, let's say EURUSD H4. What if we use not just indicators, instead, we only use correlation technique between one pair into another, for exampe EURUSD H4 with GBPUSD M15? Can Metaneural work on this issue?
Regards,
Rex
Logic creates fortune..
Member Since Feb 10, 2013
8 posts
Jan 07, 2015 at 13:47
This is possible; the correlation technique would need to be programmed into an indicator and data could be collected from that using our data collection indicator that comes with the EA. In general, multiple time frames can be used at the same time when running the EA.
Kind Regards,
Kind Regards,
Jan 09, 2015 at 09:59
This is possible; the correlation technique would need to be programmed into an indicator and data could be collected from that using our data collection indicator that comes with the EA.
To create custom indicator on one pair is standard thing to be done in programming, but creating indicator based on correlation on several pairs is another thing. Can you elaborate and explain how correlation technique programmed into an indicator? I mean, what is the algorithm used for this?
In general, multiple time frames can be used at the same time when running the EA.
Did you mean when we running this EA, we can still keep opening other timeframe? Well, I think it's based on how you build your EA, right? If in the EA structure we're not determine which timeframe (or even pair) we want to use it, then it will make a chaos whenever you change your timeframe while running EA. Or do you have a different understanding of what I suggested?
Regards,
Rex
Logic creates fortune..
Member Since Feb 10, 2013
8 posts
Jan 13, 2015 at 07:53
1. To use a correlation strategy with our EA you would need to have already have an indicator based on correlation; then you can use that indicator as a base value for data collection and feed it into our EA.
2. I meant our EA can use data from multiple time frames at once.
2. I meant our EA can use data from multiple time frames at once.
Jan 13, 2015 at 19:00
(edited Jan 13, 2015 at 19:02)
Thanks for the answers. I hope I can find good indicator explaining correlation.
I have a question that suddenly comes to mind. On the web page (https://metaneural.com/hiw.html), you explained that the data collection was done in the beginning, and this means that data collection is done only once. It also means that the data used is historical data or entire OHLC already passed.
Now, when the market moves very dynamic, market movements can change even change the character of the market itself. Are you sure that the data collection is done once? What about the market dynamics? Do you not take the data that is currently in progress?
I have a question that suddenly comes to mind. On the web page (https://metaneural.com/hiw.html), you explained that the data collection was done in the beginning, and this means that data collection is done only once. It also means that the data used is historical data or entire OHLC already passed.
Now, when the market moves very dynamic, market movements can change even change the character of the market itself. Are you sure that the data collection is done once? What about the market dynamics? Do you not take the data that is currently in progress?
Logic creates fortune..
Member Since Feb 10, 2013
8 posts
Jan 14, 2015 at 16:00
The data collection is done manually but necessarily once, you can run data collection again to refresh to matrix periodically; how often depends on your testing phase.
For example, I can use 3 years of data for collection, create the neural network 'brain' structure based on that and the EA will respond with those 3 years worth of market patterns and can do so very accurately - once you've traded with that structure for a while and entries become less accurate you can retrain. However, this is rarely necessary, market patterns don't change often and once every 6 months is more than enough.
For example, I can use 3 years of data for collection, create the neural network 'brain' structure based on that and the EA will respond with those 3 years worth of market patterns and can do so very accurately - once you've traded with that structure for a while and entries become less accurate you can retrain. However, this is rarely necessary, market patterns don't change often and once every 6 months is more than enough.
Jan 21, 2015 at 05:46
Let's say I've done data collection. The next step is building the neural model, right? What kind topology did you use on your Neural Network? Did you use CANFIS Network (Fuzzy Logic) to build the neural model or other method such as Modular or Jordan/Elman Network maybe?
Is the use of one of the topology depends entirely on the indicator or trading techniques that we use?
Is the use of one of the topology depends entirely on the indicator or trading techniques that we use?
Logic creates fortune..
*Commercial use and spam will not be tolerated, and may result in account termination.
Tip: Posting an image/youtube url will automatically embed it in your post!
Tip: Type the @ sign to auto complete a username participating in this discussion.
