NeuroNirvamanEA
NeuroNirvamanEA is a neural network-based Expert Advisor that uses a trained perceptron model to generate buy and sell signals. It combines multiple technical indicator values as inputs to the neural network and executes trades based on the network's output predictions, aiming to capture trend movements with AI-driven decision making.
Open the interactive view (equity curve, symbol switch)Key facts
Backtest results (H1, about 1 year)
Not backtested yet — this EA could not be converted automatically.
Strategy analysis
The core strategy of this EA is based on a machine learning approach using a perceptron (single-layer neural network), which takes multiple technical indicator values (such as moving averages, RSI, MACD, etc.) as input features. Entry conditions are determined by the neural network output, opening long positions when the output exceeds a specific threshold and short positions when it falls below a negative threshold. Position sizing uses fixed lots or a simple balance-based proportion method, lacking dynamic position adjustment mechanisms. Exit conditions include stop loss, take profit, and closing positions when the neural network signal reverses. Money management is relatively basic, without complex equity curve control or risk budget allocation. The overall strategy concept is innovative, but the single-layer perceptron has limited representational power and struggles to capture complex nonlinear patterns in the market.
Strengths
- Uses machine learning for trading decisions, providing some adaptive capability
- Combines multiple technical indicators as input features, offering richer information dimensions than single indicators
- Has explicit stop loss and take profit settings, providing a basic risk control framework
- Good parameter configurability allows users to adjust network and trading parameters as needed
- Neural network weights can be continuously trained and updated, enabling online learning potential
Weaknesses
- Single-layer perceptron has limited representational power and cannot capture complex nonlinear market relationships
- Lacks cross-validation and overfitting protection mechanisms, making the model prone to overfitting historical data
- Does not account for the erosive impact of trading costs (spreads, slippage) on strategy returns
- Money management is overly simplistic, lacking dynamic position sizing and risk budgeting mechanisms
- Training data window selection lacks adaptivity, potentially causing significant performance variation across different market regimes
Possible improvements
- Upgrade to multi-layer neural networks or introduce deep learning frameworks to enhance model representational power
- Add cross-validation, regularization, and early stopping mechanisms to prevent overfitting
- Implement dynamic position sizing based on model confidence and market volatility
Risk assessment
Overall risk: Medium
- Single-layer perceptron model may generate unpredictable false signals, especially during market regime transitions
- Lacks protection against extreme market events (flash crashes, gaps), where stop losses may fail due to slippage
- No out-of-sample validation; the model may severely underperform in unseen market conditions
Evaluation scores
Parameters
| Parameter | Default | Description |
|---|---|---|
Lots | 0.1 | Fixed trading lot size |
StopLoss | 500 | Stop loss in points |
TakeProfit | 500 | Take profit in points |
TrailingStop | 0 | Trailing stop in points, 0 to disable |
NetworkInputs | 8 | Neural network input feature count |
LearningRate | 0.01 | Neural network learning rate |
Threshold | 0.5 | Signal trigger threshold |
MaxPositions | 1 | Maximum positions |
Timeframe | H1 | Trading timeframe |
Similar Machine Learning EAs
Backtests are historical simulations and do not guarantee future results. Not financial advice.