AegisAI Trader

AegisAI Trader - smart signals powered by logistic regression

It uses a machine learning logistic regression model to generate signals based on a combination of real-time market dynamics. It evaluates each new bar using features such as:

• Price change and momentum

• High–low range

• Volume activity

• Correlation with a second symbol

• And other meaningful price behaviors

These features are processed exactly as during training, then combined using model coefficients to calculate a trading probability. Based on this probability, the system makes decisions using configurable thresholds:

• BUY when probability exceeds the Upper Threshold

• SELL when probability falls below the Lower Threshold

• HOLD otherwise

AegisAI Trader uses a safe, fixed-position strategy with no grid, no martingale, and no averaging. Instead, it applies intelligent Stop Loss (SL) and Take Profit

(TP) levels calculated based on recent market volatility.

You can modify SL ratio and TP ratio  in the EA inputs to fit your own strategy preferences.

The EA includes inputs for 10 coefficients:

• 9 for feature weights (learned from market data)

• 1 bias (intercept)

These coefficients come from a logistic regression model trained in Python using real historical data.

You do not need to use Python or retrain the model yourself - unless you want to.

I personally retrain the AI model every month using fresh data and will provide updated coefficients for your EA.

Just copy and paste the new numbers into the settings - that’s it.

For advanced users, Python scripts are available upon request if you wish to retrain or experiment.

If you choose to retrain the model:

1. Set Generate files = true in the EA settings. It will create two data files with market history.

2. Copy those files into the ml_train Python folder.

3. Open a terminal in that folder and run the provided scripts (Python with pandas, numpy , scikit-learn is required).

4. A new logreg_coef.txt file will be created. Copy the coefficients back into the EA.

Installation instructions (PDF) and support are available.

• Symbol: EURUSD

• Timeframe: H1 (1-hour)

• For best results, use the same timeframe in training and live trading.