After a lot of exploration, perusing incredible essentials at babypips and trying different things with different frameworks, I have at long last executed a guaging framework in light of Neural Networks. Neural networks based frameworks are demonstrated in monetary determining and overall in learning examples of a non-straight frameworks.
I accept emphatically that forex market is a non-direct framework which is challenging to show. However, one beneficial thing of forex market is that it addresses a few examples which when known can be applied in going with exchanging choices. Evidence of this idea is specialized examination and speculations that are broadly involved by brokers in recognizing these examples. This makes brain networks a superior device for forex market as brain networks are know their capacity of learning obscure cycles and estimate the examples of the interaction ahead.
Lets get to the main point. In this thread I would describe the working of the system I have developed rather than the system itself as it is a long way to explain. The system is basically a time-series-forecasting system, which means we give as input information about current Bar and the system would give out information about Bars in future. I am sure now you are thinking if its really possible and if so what are the actual inputs and what are the actual outputs. I would strongly suggest the reader to go through the basics of neural networks before reading further.
An Introduction to Neural Networks forecasting
Note: The system is optimized and targeted to trade on EURUSD 15min chart
INPUTS
EMA(t) – EMA (t-1)
EMA(t-1) – EMA (t-2)
EMA(t-2) – EMA (t-3)
RSI(t)
RSI(t-1)
%R(t)
%R(t-1)
MACD(t)
MACD(t-1)
Stoch(t)
Stoch(t-1)
1H EMA(t)
1H EMA(t-1)
OUTPUTS
EMA(t) – EMA(t+1)
EMA(t+1) – EMA(t+2)
EMA(t+2) – EMA(t+3)
where t implies current time or Bar. As you see, we give the distinction of the beyond three EMA of the 15min graph Bars, past two marker upsides of RSI, %R, MACD, Stochastic and past two EMA contrast of 1H diagram Bars. The framework will actually want to figure and result the future 3 EMA distinction of 15min diagram Bars. How could this be ready to gauge is difficult to make sense of (might be I will begin a subsequent string), yet until further notice you ought to consider the framework can do as such as it has been prepared on a few past information (3-6 months). The framework until further notice comprises of a marker and an incorporate document (both connected toward the end). I would show a few models how it tends to be utilized to distinguish the examples of forex market and go with exchanging choices.
Dashed Aqua blue line (bottom most): EMA 5 bar
Yellow line: Forecasted EMA(t+1) / output 1 of neural network
Green Yellow line:Forecasted EMA(t+2) / output 2 of neural network
Gold line (top most):Forecasted EMA(t+3) / output 3 of neural network
In the figure above it is clear from the forecast lines that the trend is going to be Bullish eventhough the current EMA shows Bearish trend. This is a kind of lead indicator.
BUY SIGNAL
1) All three forecasts significantlyabove current EMA
2) RSI trending up from below 50 level and about or crossed 50 upwards
3) Stochastic Main > Signal and trending up from over sold region
4) Optional: MACD going negative to positive
Example
SELL SIGNAL
- 1. Each of the three estimates significantlybelow current EMA
- 2. RSI moving down from above50 level and about or crossed 50 downwards
- 3. Stochastic Main < Signal and moving down from overbought locale
- 4. Discretionary: MACD going positive to negative
- EXAMPLE
Installation:
Kindly read the Instructions.txt record in the compress document appended beneath. A bit by bit method to introduce the NeuroTrend is reported in the document. On the off chance that you find trouble in establishment make it a point to me……
At last I might want to say that mess around with the framework and assuming you as it do remark. Assuming you have thoughts or ideas to apply it or adjust it, share it in this string. My principal point in making it open to this discussion is to get specialists ideas and further develop it and gain from the experience of its applications…..of course additionally acquire several bucks hehehehe
Note: Due to document connection restriction, the genuine records are zoomed to NeuroTrendv1.0
