Machine Learning Grid Trading Expert Advisor

This Expert Advisor leverages neural network computations with backpropagation for continuous learning and grid trading to achieve cost averaging.

Lin Li

Features

Neural Network Integration: Utilizes a neural network to filter and execute high-probability trades, enhancing decision-making accuracy.

RSI Indicator: Employs the Relative Strength Index (RSI) for precise entry and exit signals, ensuring trades are based on robust technical analysis.

Grid Trading: Implements a grid trading strategy to cost average trades, reducing risk and optimizing entry points across a range of market conditions.

Risk Management: Features adjustable take profit and stop loss levels, along with dynamic lot sizing based on account balance. This ensures that lot sizes increase with a rising account balance and decrease when the account balance falls, maintaining a balanced risk profile.

Optimisation

The following parameters can be adjusted to optimize the EA's performance:

Neural Network Target Values:

  • Target value for predicting a long entry signal.

  • Target value for predicting a short entry signal.

Learning Rate: Controlling the speed of weight adjustments during training. A crucial parameter for fine-tuning the neural network's responsiveness and stability.

Coefficient Applied to Weights: Scale the influence of neural network weights, affecting how inputs are transformed into outputs.

Take Profit and Stop Loss Pips: Adjust levels to define the profit and loss limits for each trade.

RSI Lookback Period: Number of bars to consider for RSI calculation, which can be fine-tuned for sensitivity to market movements.

Grid Trading Parameters:

  • Distance between grid level, determine the spacing of additional trades.

  • Lot sizing for each grid level trade, aiding in cost averaging

Overbought and Oversold Levels: RSI thresholds for identifying entry and exit points, customizable to suit different market conditions.

Code Snippets:

Neural Network & Backpropagation functions

Neural Network functions

Functionality

Entry Signals:

  • Long Entry: Initiates a long position when RSIBuffer[2] <= 30 and RSIBuffer[1] < 30, indicating an oversold condition.

  • Short Entry: Initiates a short position when RSIBuffer[2] >= 70 and RSIBuffer[1] < 70, indicating an overbought condition.

Neural Network:

  • Processes RSI values through a neural network to filter out low-probability trades and only act on high-probability signals.

  • Utilizes backpropagation for continuous improvement, adjusting node weights based on prediction errors to enhance the network's accuracy over time.

Grid Trading:

  • Executes multiple trades at set intervals above and below the current bid/ask price to cost average trades that move in the opposite direction of the initial position.

Exit Signals:

  • Long Exits: RSI reaches 60 or the predefined stop loss/take profit levels are hit.

  • Short Exits: RSI reaches 40, or the redefined stop loss/take profit levels are hit.

Code Snippets

OnTick function

Key Takeaways

Combining a neural network and a grid trading system with a traditional technical indicator such as an RSI can be a good way to improve the profitability and the drawdown profile of your trading strategy.

For more information or to discuss your project, shoot me a message on Upwork!